Author + information
- Received February 18, 2004
- Revision received June 1, 2004
- Accepted June 9, 2004
- Published online September 15, 2004.
- Eugenia Nikolsky, MD, PhD*,
- Roxana Mehran, MD, FACC*,
- Amir Halkin, MD*,
- Eve D. Aymong, MD, FACC*,
- Gary S. Mintz, MD, FACC*,
- Zoran Lasic, MD, FACC*,
- Manuela Negoita, MD*,
- Martin Fahy, MSc*,
- Shoshana Krieger, BA*,
- Issam Moussa, MD, FACC*,
- Jeffrey W. Moses, MD, FACC*,
- Gregg W. Stone, MD, FACC*,
- Martin B. Leon, MD, FACC*,
- Stuart J. Pocock, PhD† and
- George Dangas, MD, PhD, FACC*,* ()
- ↵*Reprint requests and correspondence:
Dr. George Dangas, Cardiovascular Research Foundation, 55 East 59th Street, 6th Floor, New York, New York 10022
Objectives This study was designed to assess the safety of arteriotomy closure devices (ACDs) versus mechanical compression by meta-analysis in patients undergoing percutaneous transfemoral coronary procedures.
Background Although ACDs are widely applied for hemostasis after percutaneous endovascular procedures, their safety is controversial.
Methods Randomized, case-control, and cohort studies comparing access-related complications using ACDs versus mechanical compression were analyzed. The primary end point was the cumulative incidence of vascular complications, including pseudoaneurysm, arteriovenous fistula, retroperitoneal hematoma, femoral artery thrombosis, surgical vascular repair, access site infection, and blood transfusion.
Results A total of 30 studies involving 37,066 patients were identified. No difference in complication incidence between Angio-Seal and mechanical compression was revealed in the diagnostic (Dx) setting (odds ratio [OR] 1.08, 95% confidence interval [CI] 0.11 to 10.0) or percutaneous coronary interventions (PCI) (OR 0.86, 95% CI 0.65 to 1.12). Meta-analysis of randomized trials only showed a trend toward less complications using Angio-Seal in a PCI setting (OR 0.46, 95% CI 0.20 to 1.04; p = 0.062). No differences were observed regarding Perclose in either Dx (OR 1.51, 95% CI 0.24 to 9.47) or PCI (OR 1.21, 95% CI 0.94 to 1.54) setting. An increased risk in complication rates using VasoSeal in the PCI setting (OR 2.25, 95% CI 1.07 to 4.71) was found. The overall analysis favored mechanical compression over ACD (OR 1.34, 95% CI 1.01 to 1.79).
Conclusions In the setting of Dx angiography, the risk of access-site-related complications was similar for ACD compared with mechanical compression. In the setting of PCI, the rate of complications appeared higher with VasoSeal.
Arteriotomy closure devices (ACD) have emerged as an alternative to traditional mechanical compression after percutaneous coronary intervention (PCI) (1–2). These devices have the potential to reduce the time to hemostasis, facilitate patient mobilization, decrease hospital length of stay, and improve patient satisfaction (3–13). However, the issue of ACD safety with respect to vascular complications remains controversial (14). To provide an overall assessment of vascular complications, we performed a meta-analysis of published studies that evaluated the use of various ACD versus mechanical compression in patients undergoing coronary procedures with the transfemoral approach.
Literature search and study selection
The methodology of this meta-analysis was performed following the recommendations of the Cochrane Collaboration and Meta-analyses of Observational Studies in Epidemiology (MOOSE) group (15,16). To identify eligible manuscripts, a database search (Cochrane Library, MEDLINE, CINAHL, EMBASE) for literature published since 1991 was performed in April 2003 using the following keywords: “closure device,” “hemostasis,” “femoral,” “arterial,” “vascular access,” “suture,” “collagen,” “angiography,” “angioplasty,” and “heart catheterization.” In addition, the reference lists of the identified articles were critically reviewed.
Literature search and assessment of relevance of each source were performed by two independent reviewers (E.N., A.H.). Disagreements about whether a trial should be included were resolved by discussion and adjudicated by a third reviewer (G.D.). Studies considered for the meta-analysis included randomized control trials, cohort, and case-control studies that met all of the following criteria: 1) were published as full articles; 2) had well-described protocols of intervention, including a precise report on each of the major vascular complications; and 3) compared access-related complications using ACD versus mechanical compression (manual, by means of hand compression, sandbag, C-clamp, or Femostop) in patients undergoing PCI with transfemoral access.
Pooled average rate of complications plus patients’ main demographics and procedural data were extracted from each manuscript. The primary outcome measure was cumulative incidence of major vascular complications, including pseudoaneurysm requiring ultrasound-guided compression or surgical repair; arteriovenous fistula; retroperitoneal hematoma causing hemodynamic compromise, surgery, blood transfusion, prolonged hospitalization and/or death; femoral artery thrombosis (vessel occlusion requiring surgery or thrombolysis); surgical vascular repair; access-site infection necessitating treatment with antibiotics and/or surgical drainage; and blood transfusion.
Studies assessing Angio-Seal (Angio-Seal Daig, Minnetonka, Minnesota) (3,6,9,11,12,17–23), VasoSeal (VasoSeal Datascope Inc., Montvale, New Jersey) (4,5,19,24–30), and Perclose (ProStar Plus or TecStar; Perclose, Redwood City, California) (7,8,10,12,13,18,19,21,23,29,31–35) were included in the meta-analysis. Data on various devices as well as on diagnostic (Dx) catheterization and PCI were assessed separately and in combination. In addition, a separate meta-analysis for randomized trials only was performed. Studies using different ACD sheath sizes were also analyzed.
All statistical analyses were performed using the software “Comprehensive Meta-Analysis” (Biostat Inc., Englewood, New Jersey). Study effects are presented using odds ratios (OR). An odds ratio <1 indicates a better outcome for ACD, whereas an OR >1 indicates a better outcome for mechanical compression. The OR and 95% confidence intervals (CI) for combined studies were calculated using Mantel-Haenszel fixed-effects models unless evidence of between-study heterogeneity existed, in which case Mantel-Haenszel random effects models were used (36). The random effects model allows some of the variation in ORs between studies to reflect true underlying differences and leads to a wider CI (36). Heterogeneity across studies in the meta-analysis was assessed using the Q statistic (15).
A total of 81 potentially eligible studies were identified. Of these, 51 were excluded (Appendix) for one or more of the following reasons: the lack of a control arm (n = 31), ACD use in a setting other than percutaneous coronary procedure (n = 3), the lack of precise report on each of the major vascular complications (n = 8), combined analysis of vascular complications in ACD group that included >1 device (n = 3), bias in patient selection (n = 3), review article (n = 3), use of historical control (n = 2), and repeat data reporting (n = 2).
The remaining 30 studies meeting the inclusion criteria and the important demographic and procedural determinants of patient outcome are summarized in Tables 1 and 2.⇓⇓The number of enrolled patients in the studies ranged from 60 to 10,001 (8,34), with 57% of studies having ≥100 patients in each arm. In all studies, most patients were men (range 56% to 94%) (5,8). The proportion of patients with diabetes ranged from 8% to 43% (4,8). The majority of studies (73%) used a ≥8-F device sheath. There were no significant baseline differences among ACD patients and control groups in each of the included studies.
The selected studies included a total of 37,066 patients (12,596 and 24,470 patients in the device and control groups, respectively). Among the included publications, there were 18 randomized studies (2,339 and 2,197 patients) (3–11,13,24–28,31–33) and 12 nonrandomized studies (10,257 and 22,273 patients) (12,17–23,29,30,34,35). In 12 studies, the ACD compared with mechanical compression was Angio-Seal (3,6,9,11,12,17–23); in 9 studies, VasoSeal (5,6,25–30); and in 15 studies, Perclose (7,8,10,12,13,18,19,21,23,31–35). Six studies compared more than one device versus conventional compression techniques (12,18,19,21,23,29). Separate data on Dx and PCI procedures were provided in 6 (6,7,12,25,30,34) and 19 (4,6–13,17,18,20–23,25–29,32–35) studies, respectively, whereas 6 other studies presented mixed data on Dx and PCI procedures without clearly distinguishing between Dx and PCI settings (3,5,19,22,24,31).
Angio-Seal versus mechanical compression
Among 12 studies that evaluated Angio-Seal versus mechanical compression, 2 were conducted in the Dx setting, 8 in the setting of PCI, and 4 presented results in a mixed setting (both Dx and PCI) (Fig. 1).Individual study sample sizes varied from 150 to 6,262 patients (11,20). Two studies had small sample sizes (<100 patients in each arm) (11,23). Studies analyzing the use of Angio-Seal versus mechanical compression in either the Dx or PCI setting were statistically homogeneous (Fig. 1). However, significant heterogeneity in complication rates was present in studies that analyzed mixed Dx and PCI procedures, as well as all studies. Meta-analysis did not reveal a benefit for either technique in the setting of Dx (OR 0.71, 95% CI 0.22 to 2.24), PCI (OR 0.86, 95% CI 0.65 to 1.12), or mixed procedures (OR 2.04, 95% CI 0.64 to 6.45), as well as in the overall assessment (OR 1.14, 95% CI 0.72 to 1.81). Meta-analysis of randomized trials only showed a trend toward less complications using Angio-Seal in a PCI setting (OR 0.46, 95% CI 0.20 to 1.04; p = 0.062). The results of randomized studies in any setting were similar to the results of all studies included in this analysis.
VasoSeal versus mechanical compression
Among 10 studies that evaluated VasoSeal versus mechanical compression, 2 were conducted in the Dx setting, 6 in the setting of PCI, and 4 in the mixed setting (Fig. 2).Individual study sample sizes varied from 62 to 1,956 patients (5,19). The majority of studies (8 of 10) had small sample sizes. There was no statistical evidence of heterogeneity across the studies (Fig. 2). In the setting of Dx procedures, there was no significant risk in complication rates in the ACD group compared with mechanical compression (OR 1.85, 95% CI 0.15 to 22.39); however, 2 identified studies included too small a number of patients to draw conclusions. There was no significant difference in the incidence of vascular complications when studies on mixed Dx and PCI procedures were analyzed together (OR 1.86, 95% CI 0.85 to 4.06). However, meta-analysis favored conventional technique in the setting of PCI (OR 2.52, 95% CI 1.36 to 4.65) and in any setting (OR 2.27, 95% CI 1.35 to 3.80). The results of meta-analysis of randomized trials only in the setting of PCI and in any setting were close to the results of meta-analysis of all studies.
Perclose versus mechanical compression
Fifteen studies evaluated Perclose versus mechanical compression: 3 in the Dx setting, 13 in the setting of PCI, and 5 in the mixed setting (Fig. 3).Individual study sample sizes varied from 60 to 10,001 patients (8,34). Eight studies were small, representing 53.3% of all Perclose studies. Statistical evidence of heterogeneity for the complication rate was present in studies assessing Dx and mixed procedures as well as in all studies, but not in studies related to PCI (Fig. 3). No difference in the risk of complications was found in studies assessing ACD versus mechanical compression in either setting (OR 1.51, 95% CI 0.24 to 9.47 for Dx; OR 1.21, 95% CI 0.94 to 1.54 for PCI; and OR 1.37, 95% CI 0.88 to 2.14 for any setting).
Again, the results of meta-analysis of randomized trials assessed only in the setting of PCI, mixed setting, or any setting were similar to the results of meta-analysis of all studies.
Any device versus mechanical compression
When all studied devices were analyzed versus mechanical compression, statistical evidence of heterogeneity was present across studies related to 1) Dx procedures, 2) mixed Dx and PCI procedures, and 3) any procedure. There was no significant heterogeneity across randomized trials only in either setting. There was no significant difference in the risk of complications using either technique in the randomized trials. The same was also found in the meta-analysis of all studies in the Dx and PCI settings. Meta-analysis of all included studies in mixed Dx and PCI setting and all settings together favored conventional technique over ACD (Fig. 4).
Percutaneous coronary interventions using glycoprotein IIb/IIIa inhibitors
Major vascular complication rates using Angio-Seal or Perclose versus mechanical compression in patients treated with glycoprotein IIb/IIIa inhibitors were compared in three and four studies, respectively (8,21,23,29,34). There was no evidence of significant heterogeneity across these studies (p = 0.30 and 0.76, respectively). The use of either Angio-Seal or Perclose was associated with similar complication rates to mechanical compression (OR 1.17, 95% CI 0.87 to 1.58; and OR 1.05, 95% CI 0.77 to 1.44, respectively).
Percutaneous coronary interventions using different device sheath size
Of nine studies applying Perclose in the PCI setting, four studies used closure device with sheath size 6- or 7-F (7,13,33,35) whereas the device sheath size was ≥8 French in the other five studies (8,10,21,32,34). There was no evidence of heterogeneity between studies included in either group (p = 0.95 and 0.91, respectively). The use of Perclose was associated with the same risk of complications as traditional compression regardless of device sheath size (OR 0.76, 95% CI 0.27 to 2.19 for the smaller sheath; OR 1.12, 95% CI 0.81 to 1.54 for the larger sheath).
The present meta-analysis found no significant risk with respect to vascular complications between ACD and mechanical compression in the setting of diagnostic procedures; the same was applicable for the PCI setting, with the exception of the VasoSeal device, which appeared to have a disadvantage compared to mechanical compression.
Methodologic aspects of this meta-analysis
The following important issues regarding our approach to this meta-analysis should be addressed. Primarily, our meta-analysis included both randomized and observational trials because the current literature lacks large-scale randomized control trials with sample size sufficient to reveal superiority or noninferiority of ACD compared with mechanical compression. In fact, the proportion of patients from randomized trials in the current analysis was rather low: only 8% in Angio-Seal and Perclose studies, 36% in the VasoSeal studies, and 10% in all studies.
The application of meta-analysis to observational studies has been advocated when large randomized trials are not available (16). However, there are potentially more bias sources in these studies relative to randomized trials. This may make an estimation of a single outcome potentially misleading. Inclusion of small (even randomized) studies into a meta-analysis might limit power to detect difference in major complications because of relatively low incidence of this outcome measure. This was especially true in the analysis of VasoSeal: 8 of 10 studies on this device had <100 patients in each arm. In addition, the issue of publication bias may threaten the validity of any meta-analysis, especially for observational studies. Failure to include unpublished trials introduces bias toward overestimating the effectiveness of an intervention (in this case, ACD).
Given the differences of patients that may be included in randomized controlled trials (RCT)and other studies, we also performed a meta-analysis of RCT only on this subject. The results in the setting of PCI were similar to the overall results. However, given the paucity of trials in the setting of diagnostic procedures, it was not possible to perform a meta-analysis separately for RCT in this setting.
The issue of heterogeneity
Several baseline characteristics of included patients and study design details differed among the analyzed trials. These discrepancies may have contributed to the heterogeneous results. The sources of heterogeneity in ACD studies are various and are related to patient (age, gender, body mass index, the presence of diabetes, and peripheral arterial disease), operator (individual experience and learning curves for each device), and/or procedure (femoral artery puncture site, sheath size, first versus repeat procedure, level of anticoagulation before the vascular hemostasis, and the mode of adjunctive pharmacotherapy).
Finally, the different study methodologies may be an additional reason for the between-study heterogeneity. Specifically, the indications for and the timing of blood product transfusions were not specified in the included studies, but rather left to local standards of care. Diversity of methods for the diagnosis of complications (e.g., clinical only vs. ultrasound-guided approach) and different durations of follow-up are the obvious reasons for the difference in the obtained results. This is especially true with regard to assessing rates of hematomas. In the included studies, criteria for the definition of groin hematoma had a very wide range (Table 3).Taking this consideration into account, groin hematoma was not considered a primary end point.
There was no statistically significant heterogeneity in either diagnostic or PCI setting in the Angio-Seal and VasoSeal studies. As for Perclose, the studies were found to be heterogeneous only in the diagnostic setting. The following sources of heterogeneity in the latter studies were identified: patients in the study of Kahn et al. (34), favoring mechanical compression versus Perclose, were older and more frequently diabetic than patients in the study of Gerckens et al. (7), reporting similar rates of complications in both groups. The study by Duffin et al. (12), the third study in the diagnostic setting that also did not favor either mode of hemostasis, had longer duration of follow-up (30 days) than two previous studies (7,34) (in-hospital stay solely).
Remarkably, studies assessing vascular complications in the setting of mixed procedures (both diagnostic and PCI) or in any setting (diagnostic, PCI, and both) were shown to be significantly heterogeneous (except for VasoSeal trials). Such heterogeneity is probably attributed to the principal difference in main procedural characteristics (primarily different sheath size and anticoagulation regimen). Therefore, to avoid obvious heterogeneity, the proper assessment of any ACD versus traditional compression should be performed separately for the diagnostic and PCI setting, and each device, not the entire device entity, should be compared with mechanical compression.
Relationship to current literature
Our results should also be interpreted within the context of the recently published meta-analysis by Koreny et al. (14), assessing efficacy and safety of ACD in patients undergoing cardiac catheterization. Koreny et al. (14) incorporated data on about 4,000 patients from 30 randomized studies; ACD and mechanical compression were associated with similar risk of local complications. In the same publication, using intention-to-treat data solely, the risk of hematoma and pseudoaneurysm appeared to be higher using ACD.
Despite general methodologic similarity between the study by Koreny et al. and our study, several principal differences exist. First is the different approach in defining the access site-related complications. The definition given by the individual study author was used by Koreny et al. (14); the ambiguity of including hematoma was discussed in detail earlier, and inclusion of this complication in the meta-analysis is subject to the tremendous variability of hematoma definitions (and assessment) among the various studies. Second, no separate analysis on diagnostic and interventional procedures was presented by Koreny et al. (14); vascular complications are known to depend primarily on the procedure type, and the anticoagulation and sheath size differ a great deal between diagnostic and interventional procedures (1,2). Third, 12 of 30 studies analyzed by Koreny et al. (14) were reported in abstract form, which is not considered peer-reviewed publication. Though there is no consensus regarding this issue, we chose to include only published studies providing detailed information on methodology, end point definitions, and outcomes.
Other criteria for ACD assessment
In our study, we focused principally on the comparison of complication rates using ACD versus mechanical compression for hemostasis because this matter has been in debate (14). Other important aspects in the assessment of ACD include efficacy of the specific device, time to hemostasis, ambulation and discharge, patient viewpoint, and cost-effectiveness. The analyzed studies had a wide range of ACD failure (up to 20%); this is certainly an important consideration in the overall assessment of ACD, and it understandably improves with newer generations of devices and with operator experience.
The majority of the included studies demonstrated a reduced time to hemostasis and ambulation in both the diagnostic and PCI settings using ACD versus mechanical compression (3–10,12,13,23–25,28,31–33,35); the critical issue has been a potential tradeoff of safety with ACD use, as we suggested in our previous study with the very early ACD generations (20) and another recent meta-analysis has speculated (14). According to our present meta-analysis, no such concern appears justifiable for most ACD types.
Only a few studies have assessed patient viewpoint using a vascular closure device; this is because little dispute exists on the fact that ACD application does improve patient satisfaction and time to ambulation (8,10–13,30,32). Indeed, the majority of studies have reported that patient satisfaction was higher using ACD (Angio-Seal and Perclose) than mechanical compression (8,10–13,32); however, two studies showed no difference in the level of patient satisfaction using ACD (Perclose and VasoSeal) compared with manual compression (12,30). Only two studies demonstrated a reduction of in-hospital stay (8,10) and costs (8,13) using Perclose compared with mechanical compression.
Because of the specifics of the procedure to achieve hemostasis, blinding is not possible even in randomized trials of ACD. Relatively small percentages of women were included in the trials. This meta-analysis still covered publications mostly related to the early generations of ACD. Rapidly advancing device technologies may change the position of ACD compared to the conventional compression technique. Additional studies are required to examine the safety of ACD and the impact of generational advances of these devices on outcomes.
In the setting of Dx cardiac catheterization, the risk of vascular complications related to arterial access site was similar with all three devices when compared with mechanical compression. In the setting of PCI, Angio-Seal and Perclose were similar to mechanical compression. The rate of complications after PCI was higher with VasoSeal compared to mechanical compression. Despite inherent bias and differences in study designs, our meta-analysis may provide a helpful tool to understand the sources of variability in results across studies and to improve the design of future studies on ACD.
The authors are grateful to John Zuccardy for his assistance in preparing this manuscript.
For a list of the studies excluded from the meta-analysis, please see the September 15, 2004, issue of JACCat www.cardiosource.com/jacc.html
The research fellowship program of the Cardiovascular Research Foundation is a recipient of an educational grant from St. Jude Medical Inc.
- Abbreviations and acronyms
- arteriotomy closure devices
- confidence interval
- odds ratio
- percutaneous coronary intervention
- randomized controlled trial
- Received February 18, 2004.
- Revision received June 1, 2004.
- Accepted June 9, 2004.
- American College of Cardiology Foundation
- Kussmaul W.G. 3rd.,
- Buchbinder M.,
- Whitlow P.L,
- et al.
- Gwechenberger M.,
- Katzenschlager R.,
- Heinz G.,
- Gottsauner-Wolf M.,
- Probst P.
- Nasu K.,
- Tsuchikane E.,
- Sumitsuji S.
- Duffin D.C.,
- Muhlestein J.B.,
- Allisson S.B,
- et al.
- ↵Clarke M, Oxman AD, editors. Cochrane Reviewer's Handbook 4.1.6 (updated January 2003). In: The Cochrane Library, Issue 1, 2003. Oxford: Update Software.
- Dangas G.,
- Mehran R.,
- Kokolis S,
- et al.
- Applegate R.J.,
- Grabarczyk M.A.,
- Little W.C,
- et al.
- Sanborn T.A.,
- Gibbs H.H.,
- Brinker J.A.,
- Knopf W.D.,
- Kosinski E.J.,
- Roubin G.S.
- Camenzind E.,
- Grossholz M.,
- Urban P.,
- Dorsaz P.A.,
- Didier D.,
- Meier B.
- von Hoch F.,
- Neumann F.J.,
- Theiss W.,
- Kastrati A.,
- Schomig A.
- Schickel S.I.,
- Adkisson P.,
- Miracle V.,
- Cronin S.N.