Author + information
- Received September 27, 2016
- Revision received February 8, 2017
- Accepted February 28, 2017
- Published online May 1, 2017.
- Partha Sardar, MDa,
- Saurav Chatterjee, MDb,
- Herbert D. Aronow, MDc,
- Amartya Kundu, MDd,
- Preethi Ramchand, MDe,
- Debabrata Mukherjee, MDf,
- Ramez Nairooz, MDg,
- William A. Gray, MDh,
- Christopher J. White, MDi,
- Michael R. Jaff, DOj,
- Kenneth Rosenfield, MDj and
- Jay Giri, MDk,l,∗ ()
- aDivision of Cardiovascular Medicine, University of Utah, Salt Lake City, Utah
- bSt. Luke's-Roosevelt Hospital of the Mount Sinai Health System, New York, New York
- cCardiovascular Institute, Warren Alpert Medical School at Brown University, Providence, Rhode Island
- dDepartment of Medicine, University of Massachusetts Medical School, Worcester, Massachusetts
- eDepartment of Neurology, University of Pennsylvania, Philadelphia, Pennsylvania
- fTexas Tech University Health Sciences Center, El Paso, Texas
- gUniversity of Arkansas for Medical Sciences, Little Rock, Arkansas
- hMain Line Health System, Philadelphia, Pennsylvania
- iJohn Ochsner Heart and Vascular Institute, Ochsner Clinical School of the University of Queensland, Ochsner Medical Center, New Orleans, Louisiana
- jPaul and Phyllis Fireman Vascular Center, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts
- kPenn Cardiovascular Outcomes, Quality and Evaluative Research Center, University of Pennsylvania, Philadelphia, Pennsylvania
- lCardiovascular Medicine Division, University of Pennsylvania, Philadelphia, Pennsylvania
- ↵∗Address for correspondence:
Dr. Jay Giri, Hospital of the University of Pennsylvania, Cardiovascular Medicine Division, Perelman Center, 11th Floor, 3400 Civic Center Boulevard, Philadelphia, Pennsylvania 19104.
Background Data conflict regarding the relative effectiveness of carotid artery stenting (CAS) and carotid artery endarterectomy (CEA) for the prevention of stroke due to carotid artery stenosis.
Objectives The authors performed an updated meta-analysis evaluating the efficacy and safety of CAS versus CEA, given recently published clinical trial data.
Methods Databases were searched through April 30, 2016. Randomized trials with ≥50 patients, that had exclusive use of embolic-protection devices, and that compared CAS against CEA for the treatment of carotid artery stenosis were selected. We calculated summary odds ratios (ORs) and 95% confidence intervals (CIs) using a random-effects model.
Results We analyzed 6,526 patients from 5 trials with a mean follow-up of 5.3 years. The composite outcome of periprocedural death, stroke, myocardial infarction (MI), or nonperiprocedural ipsilateral stroke was not significantly different between therapies (OR: 1.22; 95% CI: 0.94 to 1.59). The risk of any periprocedural stroke plus nonperiprocedural ipsilateral stroke was higher with CAS (OR: 1.50; 95% CI: 1.22 to 1.84). The risk of higher stroke with CAS was mostly attributed to periprocedural minor stroke (OR: 2.43; 95% CI: 1.71 to 3.46). CAS was associated with significantly lower risk of periprocedural MI (OR: 0.45; 95% CI: 0.27 to 0.75); cranial nerve palsy (OR: 0.07; 95% CI: 0.04 to 0.14); and the composite outcome of death, stroke, MI, or cranial nerve palsy during the periprocedural period (OR: 0.75; 95% CI: 0.60 to 0.93).
Conclusions CAS and CEA were associated with similar rates of a composite of periprocedural death, stroke, MI, or nonperiprocedural ipsilateral stroke. The risk of long-term overall stroke was significantly higher with CAS, and was mostly attributed to periprocedural minor stroke. CAS was associated with lower rates of periprocedural MI and cranial nerve palsy than CEA.
Carotid stenosis is responsible for almost 20% of strokes in the adult population. Stroke affects nearly 800,000 individuals in the United States per year, results in significant death and disability, and costs more than $41 billion (1–3). Treatment of carotid stenosis decreases the risk of stroke and reduces stroke-related morbidity and mortality (1–6). Carotid artery stenting (CAS) was developed in the 1990s for the management of patients with carotid stenosis and has emerged as an alternative to carotid endarterectomy (CEA) in the decades since (7–9). However, the long-term efficacy and safety of CAS relative to CEA remains controversial (10–18). Additionally, the majority of patients undergoing carotid revascularization procedures in the United States have asymptomatic carotid stenosis. Data regarding the efficacy of CAS in asymptomatic patients is limited and has not been fully evaluated in prior meta-analyses. With new data recently reported from 2 large randomized controlled trials (RCTs), we performed an updated meta-analysis of RCTs to evaluate periprocedural safety and long-term clinical outcomes with CAS in comparison to CEA for the prevention of stroke due to carotid artery stenosis.
Data sources and searches
We searched PubMed, Cochrane Library, EMBASE, EBSCO, Web of Science, and CINAHL databases from inception until April 30, 2016, for all RCTs that compared CAS with CEA. Search keywords were: carotid artery disease, endarterectomy, and carotid artery stenting. We checked the reference lists of original studies, review papers, and meta-analyses identified by the electronic searches to find other eligible trials.
Studies eligible for inclusion were RCTs that satisfied the following pre-specified criteria: 1) the trial compared CAS with CEA among patients with symptomatic or asymptomatic carotid stenosis; 2) the trial randomized ≥50 patients; 3) the trial-mandated embolic-protection devices (EPDs) use (defined as EPD use in >50% cases); and 4) the trial was a full-text English language paper published in a peer-reviewed journal. The minimum sample size requirement was included to decrease the potential effects of publication bias. We excluded studies for which only abstracts from conference proceedings were available as well as studies that examined isolated balloon angioplasty (defined as >50% of the cohort receiving isolated angioplasty). We contacted the authors of potentially relevant RCTs to clarify ambiguities on eligibility and to request relevant unpublished data. The PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) statement was followed (19).
Data extraction and quality assessment
Three authors (P.S., S.C., and A.K.) independently assessed trial eligibility, extracted data, and evaluated the risk of bias of individual trials. Disagreements were resolved by consensus among all authors. Data was extracted using an agreed data extraction form. The risk of bias was assessed using the Cochrane Collaboration tool (20).
Data synthesis and analysis
Baseline demographics, trial design, and outcomes data were extracted from each trial. The aggregate combined efficacy/safety outcome was the composite of death, stroke in any territory (ipsilateral or contralateral, major or minor), or myocardial infarction (MI) during the periprocedural period, or nonperiprocedural ipsilateral stroke. The overall efficacy outcome was the occurrence of any periprocedural stroke plus nonperiprocedural ipsilateral stroke. The long-term efficacy outcome was specified as long-term nonperiprocedural ipsilateral stroke (as defined in CREST [Carotid Revascularization Endarterectomy Versus Stenting Trial], the largest randomized prospective study comparing CAS and CEA). This referred specifically to ipsilateral strokes that occurred after the periprocedural period among patients who had no periprocedural event. The composite safety outcome was the composite of death, stroke, MI, or cranial nerve palsy during the periprocedural period. Other outcomes of interest were the individual outcomes of periprocedural death, any stroke, major or disabling stroke, minor (nondisabling) stroke, MI, cranial nerve palsy, and large neck hematoma. Vascular access complications were poorly reported in most trials and unavailable for full analysis. Long-term outcomes of interest were stroke in any territory, major or disabling stroke, minor (nondisabling) stroke, and death. Individual outcome definitions were followed as defined in each individual trial.
The intention-to-treat principle was followed whenever applicable, and the longest available follow-up data from individual trials were used. We used a random-effects model and calculated the summary odds ratio (OR) with 95% confidence intervals (CIs) (21). A 2-tailed alpha level of 0.05 was set as the threshold for statistical significance. Heterogeneity across trials was identified using I2 statistics, considering I2 <25% as low and I2 >75% as high heterogeneity, and Cochran’s Q (p ≤ 0.1) as significant for each outcome. The number needed to treat (NNT) or number needed to harm (NNH) was calculated using the following formula: NNT/NNH = 1/CER − EER, where CER indicates control event rate and EER indicates experimental event rate. We considered CEA the control group and CAS the experimental group. The analysis was performed using standard software, RevMan version 5.3 (Nordic Cochrane Centre, Cochrane Collaboration, Copenhagen, Denmark) and Stata version 12.0 (Stata Corporation, College Station, Texas). There was no funding source for this study. Methods for additional analyses are included in the Online Appendix.
Description of trials
We identified 2,405 reports from our search strategy (Online Figure 1). After complete evaluation, a total of 5 RCTs were included in the final analysis (10–18). Baseline characteristics and details of the included trials are reported in Table 1. Studies and abstracts that did not satisfy our inclusion criteria were excluded and are listed in the Online Appendix. The 5 trials included 6,526 participants, with 3,636 patients randomized to the CAS group and 2,890 patients randomized to the CEA group. Two RCTs were restricted to symptomatic patients (EVA-3S [Endarterectomy Versus Angioplasty in Patients with Symptomatic Severe Carotid Stenosis] and ICSS [Endarterectomy Versus Angioplasty in Patients with Symptomatic Severe Carotid Stenosis] trials), 2 RCTs enrolled both symptomatic and asymptomatic patients (CREST and SAPPHIRE [Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy] trials), and 1 RCT enrolled exclusively asymptomatic patients (ACT I [Asymptomatic Carotid Trial]).
The mean age of the patients ranged from 67.7 to 72.6 years, the male population varied from 56.9% to 78%, and median follow-up durations ranged from 3 to 7.4 years. A total of 4 of 5 RCTs started their recruitment nearly simultaneously (between 2000 and 2001), with 1 study initiated during 2005 (Table 1). Relevant inclusion/exclusion criteria of the trials are shown in Online Table 1. Other baseline characteristics of enrolled patients are shown in Online Table 2. The definitions of major stroke, minor stroke, and MI are presented in Table 2.
The aggregated efficacy/safety outcome (composite outcome of periprocedural death, stroke, MI, or nonperiprocedural ipsilateral stroke) was not significantly different between CAS versus CEA (295 of 3,636 [8.1%] vs. 218 of 2,890 [7.5%]; OR: 1.22; 95% CI: 0.94 to 1.59) (Central Illustration). The risk of any periprocedural stroke plus nonperiprocedural ipsilateral stroke was significantly greater with CAS (275 of 3,636 [7.6%] vs. 161 of 2,890 [5.6%]; OR: 1.50; 95% CI: 1.22 to 1.84; NNH = 50) (Figure 1). We were unable to perform a pooled landmark analysis for the occurrence of long-term nonperiprocedural ipsilateral stroke due to inadequacies in reported data in some of the published papers. Available reported data on long-term nonperiprocedural ipsilateral stroke is presented in Online Table 3. A total of 4 trials reported data on this outcome, and these individual trials all reported no significant differences in the rate of nonperiprocedural ipsilateral stroke over the longest available follow-up between the CAS and CEA groups.
Periprocedural safety outcomes
During the periprocedural period, defined in most studies as the 30 days after the intervention, the risk of the composite safety outcome (composite of death, stroke, MI, or cranial nerve palsy during the periprocedural period) was significantly lower with CAS (224 of 3,636 [6.2%] vs. 263 of 2,890 [9.1%]; OR: 0.75; 95% CI: 0.60 to 0.93; NNT = 34) (Figure 2A). Associated periprocedural stroke was significantly higher with CAS (169 of 3,636 [4.6%] vs. 73 of 2,890 [2.5%]; OR: 2.07; 95% CI: 1.56 to 2.75, NNH = 47) (Figure 2B), which was mainly attributable to minor strokes (124 of 3,636 [3.4%] vs. 44 of 2,890 [1.5%]; OR: 2.43; 95% CI: 1.71 to 3.46, NNH = 52) (Online Figure 2). There was significant heterogeneity in the definitions of disabling or major strokes in the included trials, and the data was not consistently reported in all trials (Table 2). We pooled these data for hypothesis generation, and found that the risk of associated periprocedural major stroke was not different between CAS and CEA (39 of 3,636 [1.0%] vs. 27 of 2,890 [0.9%]; OR: 1.24; 95% CI: 0.74 to 2.08) (Online Figure 3).
Patients in the CAS group experienced a similar associated periprocedural mortality rate as those undergoing CEA (26 of 3,636 [0.7%] vs. 16 of 2,890 [0.5%]; OR: 1.34; 95% CI: 0.60 to 3.02) (Online Figure 4). CAS was associated with a decreased risk of periprocedural MI (24 of 3,636 [0.6%] vs. 48 of 2,890 [1.6%]; OR: 0.45; 95% CI: 0.27 to 0.75; NNT = 99) (Figure 2C), cranial nerve palsy (9 of 3,636 [0.2%] vs. 135 of 2,890 [4.7%]; OR: 0.07; 95% CI: 0.04 to 0.14; NNT = 22) (Online Figure 5), and neurological injury (stroke plus cranial nerve injury) (178 of 3,636 [4.9%] vs. 208 of 2,890 [7.2%]; OR: 0.75; 95% CI: 0.60 to 0.95; NNT = 43) (Online Figure 6). Detailed outcome reporting is available in Table 3.
In our analysis, mean long-term follow-up was 5.3 years. The total long-term associated stroke in any territory, accounting for events both during the periprocedural and long-term follow-up period, was significantly higher with CAS (305 of 3,636 [8.4%] vs. 200 of 2,890 [6.9%]; OR: 1.47; 95% CI: 1.20 to 1.79; NNH = 68) (Online Figure 7). Again, trial-specific results for major and minor strokes were inconsistently reported and not adjudicated after the periprocedural period in some trials, and significant heterogeneity in the definitions of disabling or major strokes were present (Table 2, Online Table 4). The pooled hypothesis-generating analysis demonstrated no significant difference in associated long-term major or disabling stroke (104 of 3,636 [2.9%] vs. 89 of 2,890 [3.0%]; OR: 1.14; 95% CI: 0.76 to 1.73) (Online Figure 8). The associated risk of long-term mortality was similar between the 2 groups (429 of 3,636 [11.8%] vs. 357 of 2,890 [12.3%]; OR: 1.12; 95% CI: 0.95 to 1.31) (Online Figure 9).
CAS was associated with a lower risk of large neck hematoma (20 of 3,469 [0.6%] vs. 53 of 2,723 [1.9%]; OR: 0.33; 95% CI: 0.20 to 0.56; NNT: 73] (Online Figure 10). Available data also suggest that CAS may have been associated with a greater risk of bradycardia, hypotension, and restenosis; however, the data was inconsistently reported precluding full formal analysis (Online Table 5).
Analysis stratified by symptomatic status
We performed separate analyses for 3,654 symptomatic and 2,871 asymptomatic patients with data available from 4 and 3 trials, respectively. The composite outcome of death, stroke, or MI during the periprocedural period and ipsilateral stroke during long-term follow-up was significantly higher with CAS among symptomatic patients (210 of 1,836 [11.4%] vs. 150 of 1,818 [8.3%]; OR: 1.43; 95% CI: 1.15 to 1.79) but not among asymptomatic patients (108 of 1,800 [6.0%] vs. 88 of 1,071 [8.2%]; OR: 0.92; 95% CI: 0.68 to 1.26). With exclusion of the ICSS trial (70% use of EPD), rates of this composite outcome among symptomatic patients were no longer statistically different between CAS and CEA (115 of 983 [11.6%] vs. 88 of 961 [9.2%]; OR: 1.31; 95% CI: 0.98 to 1.76]), although a similar trend toward greater adverse events with CAS remained evident. Full results for all available outcomes stratified by symptomatic status are available in Online Figures 11 to 15.
Fixed-effects analyses showed consistent results with our primary random effect analyses. Meta-regression with multiple covariates (as mentioned before) did not detect any confounding factors/effect modifiers for the aggregate efficacy/safety outcome (composite of death, stroke, or MI during the periprocedural period and ipsilateral stroke during long-term follow-up) (Online Figure 16).
Quality assessment and risk of bias
There was no evidence of small-study effects (publication bias) by visual inspection of funnel plots (Online Figure 17) and by Egger’s regression test. Three trials were discontinued before their completion: 1 because of safety and futility (EVA3S), and 2 because of slow enrollment (SAPPHIRE and ACT1). Early termination of these trials may have influenced assessment of true effect sizes. Overall, the included trials were high-quality randomized trials. A formal risk of bias assessment is shown in Online Table 6.
Our analysis, which included data from 5 RCTs and 6,526 patients, demonstrated that the aggregate efficacy/safety outcome of death, stroke, or MI during the periprocedural period or nonperiprocedural ipsilateral stroke did not differ significantly between CAS and CEA. However, particular risks vary across the 2 procedures during the periprocedural period: minor (nondisabling) stroke is more common with CAS, whereas MI and cranial nerve palsy are more common with CEA. The risk of any stroke from the date of the procedure through long-term follow-up was significantly higher with CAS, which was mostly attributed to periprocedural minor stroke. Although we could not perform a true pooled landmark analysis, individual trial data suggest there are similar rates of long-term ipsilateral stroke outside of the periprocedural period with CAS and CEA.
Our analysis confirmed findings from multiple trials demonstrating that stenting has the disadvantage of causing more minor (nondisabling) strokes in the periprocedural period than CEA. A post hoc analysis of the CREST trial demonstrated that periprocedural minor stroke had significant effects on the physical and mental health components of the SF-36 (36-Item Short Form Health Survey) quality-of-life scale measured at 1 year (22). This drawback of CAS must be weighed against the increased risks of periprocedural MI and cranial nerve palsy associated with CEA. In the CREST post-hoc analysis, periprocedural MI was associated with decrements in the physical health component of SF-36, although to a lesser degree than minor stroke. However, periprocedural MI in CREST was associated with a 3.67-fold increase in mortality over 4 years of follow-up (23). There was no association of increased mortality with periprocedural stroke (22,23). Other observational reports have also suggested higher short- and long-term death rates after CEA in patients with procedure-related MI (24,25).
Persistence of cranial nerve palsy has been reported in 3% to 23% cases after CEA (26,27). Prior reports have demonstrated that cranial nerve palsy may be more disabling than minor stroke (26,27). These reports have raised concerns that the traditional health-related quality-of-life scales used in CAS studies may be insensitive to the degree of disability caused by cranial nerve palsy (22). However, cranial nerve palsy is transient in many instances, and the included trials did not clearly report the incidence of transient and permanent cranial nerve palsy separately. The current effort represents the first attempt to fully quantify cranial nerve palsy rates across randomized trials of CAS versus CEA. This may be particularly relevant as we discovered that rates of all periprocedural neurological injury (periprocedural stroke + cranial nerve palsy) were greater with CEA than CAS.
Technological advances in CAS, such as the development of proximal embolic protection and mesh-covered stents, aim to reduce the incidence of periprocedural stroke with CAS, but have not yet undergone rigorous comparative effectiveness analyses (3–4,28). For endarterectomy, advances in pre-operative cardiac evaluation, anesthesia, and quality improvement through standardized community-based outcome analysis are areas of focus to reduce the risk of perioperative complications (2–3,28).
Our analysis of asymptomatic patients closely mirrored the results in our overall cohort, with similar composite efficacy and safety outcome, but competing periprocedural risks including elevated periprocedural stroke risks with CAS. However, we found a higher risk of our composite efficacy/safety outcome with CAS in the symptomatic subgroup. Importantly, sensitivity analysis testing with exclusion of the ICSS trial, which used EPD in only 70% of participants, did not greatly affect the absolute risk differences seen among symptomatic patients, although this exclusion diminished our statistical power to show a difference between groups.
Prior meta-analyses comparing CAS and CEA reported higher risks of composite endpoints with CAS (1–2,4). However, most of the meta-analyses reported only short- or intermediate-term outcomes and did not fully account for all competing procedural risks, including MI and cranial nerve palsy. Our results may differ from prior analyses due to inclusion of high-quality RCTs with CAS (excluding angioplasty trials and trials with inconsistent use of EPDs), inclusion of recently reported results from 2 large RCTs (ACT-1 and CREST), inclusion of long-term data, the requirement for minimum operator experience to perform CAS in many trials, and a more exhaustive reporting and analysis of all recorded clinical trial outcomes.
Current multispecialty guidelines and existing evidence support the use of carotid revascularization for treatment of carotid stenosis in a select group of symptomatic and asymptomatic patients (3,29). The current analysis supports clinical equipoise for CAS and CEA as procedures that should be complementary, rather than competitive, therapies tailored for individual patients based on individual characteristics and preferences.
Our results are subject to the limitations inherent to meta-analyses involving the pooling of data from different trials with different study protocols, definitions of clinical outcomes, and baseline characteristics of patients. We have performed several subgroup, sensitivity, and meta-regression analyses to address these issues. There was a wide variability in the risk profile and lesion complexity of included patients, as well as variation in EPD use, types of stents used, and minimum operator experience required for participation. Included trials used several different stents/EPDs as well as varying operative techniques, and we were unable to conduct subgroup analyses by device/operation type or by all patient characteristics. All of the trials did not report data on each outcome. The subgroup analyses might suffer from biases related to multiple testing. The results of the sensitivity analyses are best described as secondary and hypothesis-generating only. Most included trials did not routinely assess periprocedural cardiac biomarkers or electrocardiograms; hence, the periprocedural MI rate may be underestimated. The trials may underestimate total long-term stroke risk due to outcome reporting that may bias away from ascertainment of recurrent strokes (i.e., landmark analyses that censor patients with a periprocedural stroke). It is unclear whether this potential ascertainment bias is differential between therapies. As this study used only published data, we could not explore results using individual, patient-level data. Finally, reported results for all comparisons that do not show statistical differences between groups should be interpreted in the context of testing a noninferiority hypothesis. Hence, to consider the groups “equal” for a given result, one must accept noninferiority margins that are as large as the upper limit of the reported CIs. For example, for our primary efficacy/safety endpoint, a statement of “equivalent” results between CAS and CEA could only formally apply if the accepted noninferiority margin between therapies was as large as 1.60.
CAS and CEA were associated with similar rates of a composite outcome of periprocedural death, stroke, MI, or nonperiprocedural ipsilateral stroke. The risk of any stroke over long-term follow-up was significantly higher with CAS, which was mostly attributed to periprocedural minor stroke. CAS was associated with lower periprocedural rates of MI and cranial nerve palsy than CEA.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: CAS and CEA are associated with similar rates of the composite outcome of periprocedural death, stroke, MI, or subsequent ipsilateral stroke. Each procedure is associated with specific safety concerns; CAS carries a greater risk of periprocedural stroke, and CEA a greater risk of MI and cranial nerve palsy.
TRANSLATIONAL OUTLOOK: Further studies are needed to define the clinical characteristics most pertinent to the selection of optimum revascularization strategies for stroke prevention in patients with symptomatic or asymptomatic carotid artery stenosis.
For an expanded Methods section as well as supplemental figures and tables, please see the online version of this article.
Dr. Gray has served as a consultant to Abbott Vascular, Medtronic, W.L. Gore, and Boston Scientific. Dr. Jaff is a noncompensated advisor for Abbott Vascular, Boston Scientific, Cordis Corporation, Cardinal Health, and Medtronic Vascular; and a compensated board member of VIVA Physicians, a 501c3 not-for-profit education and research organization. Dr. Rosenfield is a compensated advisor for Abbott Vascular, Cardinal Health, Surmodics, Inari Medical, Volcano/Phillips, and Proteon; has received stock options for serving on advisory boards from Contego, Access Vascular, and MD Insider; holds stock in Embolitech, Janacare, Primacea, and PQ Bypass; has received grant support to his institution from Abbott Vascular, Atrium/Maquet, and Lutonix/Bard; has served as a consultant to Capture Vascular, Shockwave, and the University of Maryland; has received research grant support from Inari Medical and the National Institutes of Health; has received equity or stock options from Capture Vascular, VORTEX, Micell, shockwave, Cruzar Systems, Endospan, Eximo, Valcare, and Contego; has received an honorarium from Cook; and is a compensated board member of VIVA Physicians, a 501c3 not-for-profit education and research organization. Dr. Giri has received a research grant to his institution from St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- carotid artery stenting
- carotid artery endarterectomy
- confidence interval
- embolic protection device
- myocardial infarction
- number needed to harm
- number needed to treat
- odds ratio
- randomized controlled trial
- Received September 27, 2016.
- Revision received February 8, 2017.
- Accepted February 28, 2017.
- 2017 American College of Cardiology Foundation
- Meier P.,
- Knapp G.,
- Tamhane U.,
- Chaturvedi S.,
- Gurm H.S.
- Vincent S.,
- Eberg M.,
- Eisenberg M.J.,
- Filion K.B.
- Gurm H.S.,
- Nallamothu B.K.,
- Yadav J.
- Gray W.A.,
- Chaturvedi S.,
- Verta P.
- Gray W.A.,
- Yadav J.S.,
- Verta P.,
- et al.,
- for the CAPTURE Trial Collaborators
- Mas J.L.,
- Arquizan C.,
- Calvet D.,
- et al.
- ↵Higgins J, Altman D, Sterne J. Cochrane Handbook for Systematic Reviews of Interventions. Available at: http://handbook.cochrane.org/. Accessed May 28, 2016.
- Cohen D.J.,
- Stolker J.M.,
- Wang K.,
- et al.,
- for the CREST Investigators
- Blackshear J.L.,
- Cutlip D.E.,
- Roubin G.S.,
- et al.
- Landesberg G.,
- Shatz V.,
- Akopnik I.,
- et al.
- Boulanger M.,
- Camelière L.,
- Felgueiras R.,
- et al.
- Hye R.J.,
- Mackey A.,
- Hill M.D.,
- et al.
- Brott T.G.,
- Halperin J.L.,
- Abbara S.,
- et al.