A Meta-Analysis of Percutaneous Vascular Closure Devices After Diagnostic Catheterization and Percutaneous Coronary Interventio

ABSTRACT: Background. While each of the available femoral arterial sealing devices claims equivalent safety to manual compression following invasive cardiac procedures, these claims are based on small, underpowered studies. Our aim was to increase the ability to detect a clinically meaningful difference by performing a meta-analysis. Methods. We identified studies via Medline and manual searches and selected studies that were prospective, randomized clinical trials for inclusion. Pooling of data was performed by calculation of the Mantel-Haenszel odds ratio (OR) and the variance of the OR was estimated using the method of Robins, Greenland and Breslow. Results. Sixteen studies enrolling 5,048 patients were included in the analysis. The pooled OR was 0.89 (95% confidence interval, 0.86–0.91), indicating a significant decrease in risk by devices. Excluding hematomas from the endpoint resulted in a concordant result. Angio-Seal was associated with a significant reduction in risk (OR, 0.51) and Perclose had a neutral result (OR, 1.0), whereas Vasoseal had an increased risk of complications (OR, 1.18). Conclusion. Overall, sealing devices may be associated with a reduction in risk of complications following invasive coronary procedures, but significant differences may exist among individual devices. These potential differences need to be explored in randomized, controlled clinical trials. J INVAS CARDIOL 2004;16:243–246 Key words: complications, meta-analysis, vascular sealing devices The management of arterial access sites for cardiac catheterization and percutaneous coronary interventions (PCI) continues to evolve. Recent years have witnessed the introduction of several hemostatic devices which continue to gain in popularity over the traditional approach of manual compression. Most studies evaluating these devices have been relatively small and have offered the conclusion that the devices present a safety profile comparable to manual compression and offer advantages in terms of early ambulation and patient comfort.^1–20 Because of their small size, however, these studies are uniformly underpowered to detect as significant modest but clinically significant safety advantages or disadvantages. Accordingly, a meta-analysis of the clinical trials of these devices will have proportionately greater power to assess the issue of safety. Methods Studies were identified via a Medline search of published papers up to April 2003, manual review of the bibliographies of the papers identified on Medline, review of abstracts presented at the meetings of the American College of Cardiology, American Heart Association, European Society of Cardiology and Transcatheter Cardiovascular Therapeutics for 1999 to March 2003, with subsequent Medline searches of the authors of published abstracts, and review of the websites of the companies involved in the manufacture and distribution of commercially available devices (www.vasoseal.com, www.vascularsolutions.com, www.perclose.com and www.sjm.com). Studies were selected for inclusion if they met the following criteria: 1) randomized, prospective, controlled clinical trial comparing a device or devices to manual compression (whether or not aided by compression devices such as the FemoStop device; 2) published as a manuscript in an English-language journal; and 3) provided data on clinical complication rates. There were 16 studies that met these criteria.^1–16 Studies were rejected if they were published only in abstract form,^18,19,22,23 were not prospective or randomized²⁰ or examined endpoints such as hemostasis time but did not provide data on clinical outcomes.¹⁷ In the case of redundant publication, the more detailed manuscript was selected. As relatively few studies presented data on individual clinical complications, we elected to examine the combined endpoint of major or serious complications as reported in the original trials. The individual elements of the combined endpoint are summarized in Table 1. The apparent heterogeneity among the trials’ definitions of major complications is not very great. Small semantic differences belie an actual close concordance. Thus, while one study may have specified “arterial occlusion” whereas another specified “thromboembolism,” in reality the two definitions would capture the same set of complications. However, to achieve as much uniformity as possible, for several studies we recalculated the “major” complication rate.^7,10 This was done because the manuscripts’ original definitions of “major” and “minor” may have been narrowly constructed, but the manuscripts presented detailed data on individual endpoints which the majority of other studies had included in their definition of “major” complication.^7,10 The single element of “major” complication that may have presented the greatest qualitative source of heterogeneity was “large” hematoma. Several studies did not include hematomas or present data on hematomas at all,^3,11 and there was little uniformity in the definition of “large” hematoma. Thus, we repeated the analysis of major complications, having deleted all cases of hematoma. This is also clinically rational because compared to the other categories of major complications, a large hematoma per se generally poses a less serious threat and requires less intervention than the other major complications. The principal analysis included all eligible studies. Secondary analyses included the stratification of studies according to the type of cardiac procedure (diagnostic catheterization versus PCI). Only studies in which these two subgroups could be clearly separately extracted were included in this analysis. Also, studies utilizing each of three closure devices (Vasoseal, Angio-Seal, Perclose) were separately meta-analyzed. We performed analyses of each of the three devices for all cases and for PCI cases only. For the Duett closure system, we located only a single published manuscript of a controlled clinical trial,¹⁵ and thus no meta-analysis could be performed for this device. We calculated the odds ratios (OR) and 95% confidence intervals (CI) for each of the individual studies. Pooling of data was performed by calculation of the Mantel-Haenszel OR and the variance of the OR was estimated using the method of Robins, Greenland and Breslow; a 95% CI was calculated accordingly.²⁴ To assess for statistical heterogeneity, we calculated a Q statistic for all pooled studies as well as each of the subsets of studies pooled for each of the individual devices.²⁵ Results Sixteen studies enrolling 5,045 patients were included in the analysis and are summarized in Table 2. Nine of the studies included only patients who had undergone PCI, two studies enrolled only patients undergoing diagnostic catheterization and the remaining 5 studies enrolled a mix of PCI and diagnostic cases. The ORs for each of the individual studies and the principal pooled analysis are illustrated in Figure 1. The pooled OR was 0.89 (95% CI, 0.86–0.91), indicating a significant decrease in risk by devices. Excluding hematomas from the endpoint resulted in a concordant result, with an OR of 0.86 (95% CI, 0.83–0.89), again indicating a significant reduction in complications with sealing devices. The Q statistic for this pool was 44.8 (p 26 I would add to this call for better reporting in publications by insisting that studies be required to provide power calculations in their methods sections and revisit the issue by stating in their discussion how large an effect the study may have missed given their final sample size. In our meta-analysis, we made every reasonable effort to render the analyses of each of the devices equal. What constitutes a major complication is a topic for debate. We attempted to achieve uniformity of this endpoint by applying adjustments to the published data. As the majority of studies included pseudoaneurysms as major complications, for those studies that chose to deem these as minor complications, we added these events to the category of major complications. It could be argued that pseudoaneurysms are simple to detect and diagnose, and are relatively simple to manage and are rarely associated with long-term morbidity. Nevertheless, even the “simple” diagnostic and management techniques consume resources and are subjectively unpleasant for the patient. The continued evolution of the devices may affect their complication rates. Thus, newer versions of these devices may have enhanced their safety and efficacy. However, in the absence of controlled clinical trials comparing newer to older devices, we cannot be certain whether the innovations, which are usually targeted to improved user friendliness, lead to improvement or worsening of clinical outcomes. It must be underscored that the present meta-analysis did not directly compare one device to another. Each device was compared to a control group managed by manual compression. The natural inclination is to examine each of these comparisons to the placebo-controlled patients and conclude that Angio-Seal and Perclose are superior to Vasoseal. While it is correct to acknowledge that heterogeneity between studies makes such inter-device comparisons hazardous at best, it remains valid to compare each device to manual compression. Within each individual controlled clinical trial, the randomization process ensured that the various elements that would impact outcomes (e.g., sheath size, type and intensity of anticoagulation) were distributed equally between treatment groups. As the use of these devices spans the spectrum of invasive practice, our expectation is that their safety results should equally be demonstrated across the entire spectrum. Like all meta-analyses, the present effort cannot be construed as definitive evidence in favor of sealing devices, but is useful in generating hypotheses and in bringing sense to a field dominated by a series of small studies. In that regard, the current meta-analysis raises questions about the conclusions of “equivalence” of the various devices versus manual compression. The most appropriate means of answering the question of whether significant differences exist among the various closure devices is to perform an appropriately powered, randomized, prospective, multi-center trial. Recent efforts have attempted to accomplish this,27 but have again enrolled too few patients to draw firm conclusions. The logistics and economics of conducting an appropriately sized comparative trial are daunting and it is unlikely that such a trial will be undertaken. Consequently, meta-analysis of existing smaller and imperfect trials will remain the approach that provides the interventional community with the greatest statistical power.