Author + information
- Received June 29, 2016
- Revision received August 29, 2016
- Accepted September 8, 2016
- Published online November 28, 2016.
- Laurent Bonello, MD, PhDa,b,c,∗ (, )
- Marc Laine, MDa,c,
- Etienne Puymirat, MD, PhDd,e,
- Gilles Lemesle, MD, PhDf,
- Franck Thuny, MD, PhDa,c,
- Franck Paganelli, MDa,
- Pierre Michelet, MD, PhDc,g,
- Antoine Roch, MD, PhDc,h,
- François Kerbaul, MD, PhDc,i and
- Laurent Boyer, MD, PhDj
- aService de Cardiologie, Centre Hospitalier Universitaire de Marseille, Hôpital Nord, Aix-Marseille Université, Marseille, France
- bAix-Marseille Université, INSERM UMR-S 1076, Vascular Research Center of Marseille, Marseille, France
- cMARS Cardio, Mediterranean Association for Research and Studies in Cardiology, Hôpital Nord, Marseille, France
- dDépartement de Cardiologie, Hôpital Européen Georges Pompidou, Assistance Publique des Hôpitaux de Paris, Paris, France
- eUniversité Paris Descartes, INSERM U-970, Paris, France
- fService d’Accueil des Urgences, Hôpital Timone, Marseille, France
- gDépartement de Cardiologie, CHU de Lille, Lille, France
- hService d’Accueil des Urgence, Hôpital Nord, Marseille, France
- iPole RUSH, Assistance–Publique Hôpitaux de Marseille, Marseille, France
- jEA 3279–Public Health, Chronic Diseases and Quality of Life–Research Unit, Aix-Marseille University, Marseille, France
- ↵∗Reprint requests and correspondence:
Dr. Laurent Bonello, Service de Cardiologie, Hôpital Universitaire Nord de Marseille, Aix-Marseille Université, INSERM UMRS 1076, Chemin des Bourrely, Marseille 13015, France.
Objectives The aim of this study was to compare an early versus a delayed invasive strategy in non–ST-segment elevation acute coronary syndromes by performing a meta-analysis of all available randomized controlled clinical trials.
Background An invasive approach is recommended to prevent death and myocardial infarction in non–ST-segment elevation acute coronary syndromes. However, the timing of angiography and the subsequent intervention, when required, remains controversial.
Methods A previous meta-analysis of 7 randomized clinical trials comparing early and delayed invasive strategies in non–ST-segment elevation acute coronary syndromes with 3 new randomized clinical trials identified in a search of the published research (n = 10 trials, n = 6,397 patients) was updated.
Results The median time between randomization and angiography ranged from 0.5 to 14.0 h in the early group and from 18.3 to 86.0 h in the delayed group. There was no difference in the primary endpoint of mortality (4% vs. 4.7%; random-effects odds ratio [OR]: 0.85; 95% confidence interval [CI]: 0.67 to 1.09; p = 0.20; I2 = 0%). The rate of myocardial infarction was also similar (6.7% vs. 7.7%; random-effects OR: 0.88; 95% CI: 0.53 to 1.45; p = 0.62; I2 = 77.5%). An early strategy was associated with a reduction in recurrent ischemia or refractory angina (3.8% vs. 5.8%; random-effects OR: 0.54; 95% CI: 0.40 to 0.74; p < 0.01; I2 = 28%) and a shorter in-hospital stay (median 112 h [interquartile range: 61 to 158 h] vs. 168 h [interquartile range: 90.3 to 192 h]; random-effects standardized mean difference −0.40; 95% CI: −0.59 to −0.21; p < 0.01; I2 = 79%). Major bleeding was similar in the 2 groups (3.9% vs. 4.2%; random-effects OR: 0.94; 95% CI: 0.73 to 1.22; p = 0.64; I2 = 0%).
Conclusions An early invasive strategy does not reduce the risk for death or myocardial infarction compared with a delayed strategy. Recurrent ischemia and length of stay were significantly reduced with an early invasive strategy.
The current guidelines for managing non–ST-segment elevation acute coronary syndromes (NSTE-ACS) recommend an invasive strategy in most patients (1). In fact, in this clinical setting, an invasive strategy reduces the risk for the composite of death and myocardial infarction (MI) compared with a selectively invasive or conservative strategy (1–3). Stabilization of the culprit lesion with antithrombotic agents and revascularization offers optimal protection from adverse events. However, the optimal timing of the invasive strategy remains controversial, and the guidelines to date have been based on limited evidence. The current European Society of Cardiology (ESC) guidelines are based on relatively old trials and a meta-analysis published in 2013 (1,4). The TIMACS (Timing of Intervention in Acute Coronary Syndromes) trial is the largest trial on this topic to date, with more than 3,000 patients included. However, that study succumbed to early termination because of difficulties in enrollment and did not meet its primary endpoint; therefore, its results should be analyzed with caution (5). A meta-analysis performed in 2013 summarized both randomized clinical trials (RCT) and observational data, comparing early invasive and delayed invasive strategies. The investigators observed nonsignificantly decreased mortality with an early invasive strategy in the RCT and a nearly significant difference in the observational studies. However, they also observed a trend toward an increased rate of MI in the early strategy group. However, overall, none of these trends reached statistical significance, and although they are hypothetical, they remain the foundation of the current guidelines (4). Since that time, new RCTs have been published comparing the early and delayed invasive strategies; these additional data may help resolve the debate on the potential benefit of each approach. We therefore aimed to provide an updated meta-analysis of prospective controlled randomized trials performed to date to determine the optimal timing of the invasive strategy in NSTE-ACS to prevent death.
We performed a meta-analysis according to the established method and in adherence with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement for reporting systematic meta-analyses in health care interventions (6). The present meta-analysis was not registered in a review registry.
We performed a publication search in MEDLINE and the Cochrane Library for RCTs published between November 1994 and February 2016 using the keywords “early coronary intervention,” “delayed coronary intervention,” “acute coronary syndrome,” “timing,” “early percutaneous coronary intervention (PCI),” “non ST-elevation acute coronary syndromes,” “non-ST elevation MI,” “coronary invasive,” “NSTE-ACS,” “NSTE MI,” “early coronary intervention,” “early coronary angioplasty,” and “ACS.” A flowchart is provided in Online Figure 1 with detailed descriptions of publication screening and reasons for exclusions.
Studies eligible for inclusion in this meta-analysis were RCTs comparing early and delayed invasive strategies in patients presenting with NSTE-ACS (7–15). We included studies in which patients were randomly allocated to early or delayed diagnostic angiography. We also included the OPTIMA (Optimal Timing of Coronary Intervention in Unstable Angina) trial, in which patients were randomized at the time of coronary angiography if they were eligible for PCI (7). We excluded studies comparing invasive and conservative management of NSTE-ACS. Accordingly, for the 3-group trial (LIPSIA-NSTEMI [Leipzig Immediate Versus Early and Late Percutaneous Coronary Intervention Trial in NSTEMI]) that compared immediate, early, and selective invasive groups, we excluded the delayed group, which was a selective invasive group. For that trial, in the primary analysis, we included the data of the very early group (<2 h) in the early invasive strategy group and of the early group (10 to 48 h) in the delayed invasive group of the meta-analysis (8).
Inclusion criteria for RCTs included a diagnosis of NSTE-ACS and allocation to an early or a delayed invasive strategy. The early intervention group of the present meta-analysis included all early invasive strategy groups as defined in each trial and the very early invasive strategy group of the LIPSIA-NSTEMI trial. The delayed invasive strategy group of the meta-analysis included all delayed invasive strategy groups as defined in each trial and the early invasive strategy group of the LIPSIA-NSTEMI (Table 1) (8). The timing of the invasive strategy defined the allocation to each group and is represented by coronary angiography. Intervention was defined as coronary revascularization either with PCI or coronary artery bypass grafting on the basis of angiographic characteristics and physicians’ clinical judgment. In the OPTIMA trial, the randomization was performed after coronary angiography between an early intervention group, in which PCI was performed immediately, compared with a delayed group in which PCI was performed between 24 and 48 h after angiography (7).
Data extraction was performed through the main report, any accompanying supplemental material, and any secondary published analyses available. The data available were limited for the study by Zhang et al. (9).
We systematically recorded the study characteristics (number of randomized patients, enrollment period, and length of follow-up), patient demographics, risk factors, and cardiac history. We also documented the medical therapies administered to patients upon admission, timing of coronary angiography, angiographic characteristics, and type of treatment. We extracted from included studies the number of events for the following outcomes: death, MI, recurrent ischemia (RI) or refractory angina (RA), major bleeding, stroke, and in-hospital length of stay (LOS). Data extraction was performed on the basis of the intention-to-treat principle. For all events, we considered the longest available follow-up period for each trial. Two authors independently extracted the data using a pre-constructed form. Disagreements were resolved by consensus and arbitration by 2 other authors.
The primary endpoint of the meta-analysis was the rate of death. Secondary endpoints included MI, major bleeding, RI or RA, and in-hospital LOS.
Definition of MI
The definition of MI was heterogeneous among the trials (Online Table 1).
Definition of major bleeding
Major bleeding was defined using TIMI (Thrombolysis in Myocardial Infarction) criteria (intracranial hemorrhage, a ≥5 g/dl decrease in hemoglobin concentration, or a ≥15% absolute decrease in hematocrit) or as defined by the trial or study investigators.
Definition of RI or RA
RI was defined in the TIMACS trial as recurrent ischemic symptoms lasting more than 5 min while the patient received optimal medical therapy (≥ 2 antiangina treatments) with documented characteristic electrocardiographic changes indicative of ischemia and requiring additional intervention (5). In the ELISA (Early or Late Intervention in Unstable Angina) 3 trial, RI was defined as recurrent chest pain associated with new or recurrent electrocardiographic abnormalities requiring urgent or repeat angiography or repeat hospitalization (10). In the RIDDLE-NSTEMI (Randomized Study of Immediate Versus Delayed Invasive Intervention in Patients With Non ST-Segment Elevation Myocardial Infarction) trial, it was defined as repeated episodes of ischemic symptoms lasting >5 min if all of the following applied: 1) the patient was on optimal medical therapy; 2) electrocardiographic changes indicative of ongoing myocardial ischemia were present; and 3) invasive intervention was required (7). For the other trials, the definition of this endpoint was not described in the reports or in the other available sources (8,12–15).
Methodological quality of the included studies
The methodological quality of the included studies was independently assessed by 2 of the authors (L. Bonello and M.L.) using a validated checklist of items for RCTs based on the Cochrane Collaboration guidelines (concealment of treatment allocation; blinding of participants, personnel, or outcome assessors; adequate assessment of incomplete outcome data; presence of selective outcome reporting; and other potential sources of bias) (16). Any discrepancy was resolved by consensus with a third reviewer (L. Boyer).
Data were analyzed according to the intention-to-treat principle. Each outcome was analyzed using the odds ratio (OR), defined as the ratio of the odds of the studied event (i.e., death, MI, RI, major bleeding) between the 2 groups, or the standardized mean difference, defined as the difference in mean outcome (i.e., LOS) between 2 groups divided by the pooled standard deviation of the measurements. The mean and standard deviation for LOS were estimated from the sample size, median, and interquartile range using an approximation method relevant for skewed data (17).
First, we performed head-to-head direct evidence, also called pairwise meta-analyses. We synthesized studies that compared early and delayed invasive strategies using fixed- or random-effects models because of clinical heterogeneity from differences in treatment types and treatment durations. The random-effects model accounts for between-study heterogeneity by weighting studies similarly (18). Heterogeneity was assessed using the I2 statistic, which represents the percentage of variance due to between-study factors rather than sampling error (19). We considered values of I2 >50% to be indicative of large heterogeneity. When possible, we used funnel plots and the Egger regression intercept (which assesses the degree of funnel plot asymmetry by the intercept from regression of standard normal deviates against precision) to estimate the risk for bias (20). Forest plots were generated to show OR or standardized mean difference with corresponding 95% confidence intervals (CIs) for each study and the overall fixed- or random-effects pooled estimate. When possible, we performed sensitivity analyses according to the following variables: year of study (before 2010 and 2010 or later), median time of the early strategy compared with the delayed strategy (<3 and ≥3 h vs. delayed strategy), and the risk for bias. Sensitivity analyses were performed to evaluate the influence of each study on estimates of effect by removing each and then assessing the effect of the removal on the effect estimate. Analyses were performed with Comprehensive Meta-Analysis version 2.0 (Biostat, Englewood, New Jersey) (21).
Second, we performed network meta-analyses or mixed treatment comparisons between 3 median times to coronary invasive strategy (i.e., <3 h, from 3 to 24 h, and >24 h) for the most complete outcomes (i.e., death, MI, RI, major bleeding, and LOS). Using a Bayesian hierarchical random-effects model with a noninformative prior hypothesis, all direct and indirect comparisons were taken into account to reach a single consistent estimate of the effect of all included treatments based on all included studies. Consistency was assessed using the node-splitting method. A value near zero indicated that the comparisons in the network were consistent. Analyses were performed with the gemtc-R package and Winbugs (22,23).
Study and patient characteristics
We screened 1,928 reports and identified 24 RCTs that were potentially relevant. After excluding 14 of them, we were left with 10 RCTs (5,7–15), 3 of which were new to this updated meta-analysis (10,11,15) (Online Figure 1).
These RCTs enrolled a total of 6,397 patients; 3,313 were randomized to early and 3,084 to delayed invasive strategies. Trial characteristics are summarized in Table 1. Additional tables summarizing patient characteristics are available in the Online Appendix (Online Tables 1 to 3). The main characteristics of patients were available for all but 1 study (9). The median time of the invasive strategy ranged from 0.5 to 14 h in the early group and from 18.3 to 86 h in the delayed intervention group.
In all but 1 study, definitive treatment included PCI, coronary artery bypass grafting, or medical therapy. In the OPTIMA trial, only patients who underwent PCI were included.
Patients assigned to the early or delayed invasive strategy were well matched. In all studies, pre-treatment with dual-antiplatelet therapy was used in both groups. The median follow-up period was 6 months.
Risk for bias
The RCTs were similar in terms of their risks for bias (Online Table 4), except for the SISCA (Stenting in Small Coronary Arteries) trial, which had a higher risk for bias (15). All studies were performed according to the intention-to-treat principle except for the SISCA trial, for which this was unclear. Patients lost to follow-up were infrequent and were described in detail. In the OPTIMA trial (7) and the trial of Zhang et al. (9), the methods for random-sequence generation and allocation concealment were unclear. ELISA (12) and RIDDLE-NSTEMI (11) were single-center studies, while the remaining RCTs were multicenter in nature. Patients and providers were not blinded to the timing and identity of interventions, a feature common to trials of coronary angiography. However, a blinded committee or a blinded outcome assessor adjudicated outcomes, except for the SISCA trial, for which the blinded assessment was unclear (15).
Primary endpoint: Mortality
The effect of an early versus delayed strategy on mortality was described in all 10 included RCTs (Figure 1). There was no significant difference between the 2 groups in terms of mortality (4% vs. 4.7%; OR: 0.85; 95% CI: 0.67 to 1.09; p = 0.20; I2 = 0%). The associated funnel plot was reasonably symmetrical, although the limited number of studies does not allow the exclusion of publication bias (Online Figure 2). The p value of the Egger’s regression intercept was > 0.05 (Table 2).
The effect of an early versus a delayed invasive strategy on MI was available for the 10 RCTs in this meta-analysis (Figure 2). There was no significant difference between the 2 groups in terms of MI (6.7% vs. 7.7%; OR: 0.88; 95% CI: 0.53 to 1.45; p = 0.62; I2 = 77.5%). The associated funnel plot was rather asymmetrical (Online Figure 3). The p value of the Egger’s regression intercept was > 0.05 (Table 2).
RI or RA
We identified 7 studies comparing the effect of an early versus a delayed strategy on RI or RA (Figure 3). The early strategy was associated with less RI than the delayed strategy (3.8% vs. 5.8%; OR: 0.54; 95% CI: 0.40 to 0.74; p < 0.01; I2 = 28.3%). The associated funnel plot was reasonably symmetrical, although the limited number of studies does not allow the exclusion of publication bias (Online Figure 4). The p value of the Egger’s regression intercept was >0.05. The difference remained significant on the basis of year of study and the median time of the early intervention (Table 2).
The rates of major bleeding complications were described in all 10 RCTs (Figure 4). There was no significant difference between the 2 groups (3.9% vs. 4.2%; OR: 0.94; 95% CI: 0.73 to 1.22; p = 0.64; I2 = 0.0%). The associated funnel plot was reasonably symmetrical, although the limited number of studies does not allow the exclusion of publication bias (Online Figure 5). The p value of the Egger’s regression intercept was >0.05. The differences remained nonsignificant on the basis of the year of study and the median time of the early strategy (Table 2).
We identified 7 studies comparing the effect of an early versus a delayed strategy on LOS (Figure 5). The early strategy was associated with a reduced LOS compared with the delayed strategy (median 112 h [interquartile range: 61 to 158 h] vs. 168 h [interquartile range: 90.3 to 192 h]; standardized mean difference = −0.40; 95% CI: −0.59 to −0.21; p < 0.01; I2 = 79%). The associated funnel plot was rather asymmetrical (Online Figure 6). The p value of the Egger’s regression intercept was > 0.05. The differences remained significant according to the year of the study (Table 2).
Sensitivity analyses, which were performed by removing each of the studies 1 at a time, indicated that no single study changed the statistical significance of the overall results.
Indirect comparisons using network meta-analyses
Death, MI, RI, and major bleeding did not differ according to the 3 median times to coronary invasive strategy, in contrast to LOS, which was lower for median time to coronary invasive strategy <3 h than for median times from 3 to 24 h and those >24 h.
The present meta-analysis of 10 RCTs suggests that an early invasive strategy does not translate into any mortality benefit compared with a delayed invasive strategy. The network analysis confirmed the lack of survival benefit with the early strategy. These results add to a previous meta-analysis by Navarese et al. (4), who found a nonsignificant survival benefit in the RCTs and a nearly significant difference in the observational studies. On the basis of the findings of the updated meta-analysis, it is unlikely that a small difference in the delay of intervention between the 2 groups of patients with NSTE-ACS monitored in an intensive care unit could translate into a mortality difference (1). In addition, our results suggest similar rates of MI with the 2 therapeutic strategies and do not confirm the trend observed by Navarese et al. of an increased rate of MI with an early strategy. The meta-analysis shows heterogeneity among studies regarding this outcome. In addition, it must be acknowledged that the definition of MI was variable between the trials and often included periprocedural MI, which has no prognostic significance (24). Finally, 1 additional difficulty in assessing MI is that in a significant number of patients, biomarkers had not returned to normal values or were still in the ascending phase of the curve at the time of randomization. Therefore, it may have been difficult, if not impossible, to differentiate between the evolution of the index MI and an ischemic complication of the revascularization procedure. Future trials on this topic should use a universal definition of MI to ensure uniform and reproducible data (25).
The only significant clinical difference observed in the present meta-analysis was reductions in RI and RA with an early invasive strategy. Such a strategy reduced the risk for RI by nearly 50%. It is important to note that both early and delayed strategies were associated with similar rates of major bleeding, confirming that an early intervention in this clinical setting is safe.
In acute coronary syndromes, the aim of the therapeutic strategy is to stabilize the culprit lesion to prevent thrombus growth and its complications (1). An invasive strategy has been shown to be beneficial over a selective invasive strategy by reducing death and MI and is therefore recommended (2,3). The early invasive strategy aims to stabilize the culprit lesion through early revascularization in conjunction with antithrombotic therapies (1). The delayed strategy aims to stabilize the culprit lesion using antithrombotic agents as a pre-treatment to facilitate delayed revascularization. Our findings suggest that this delay to prepare the culprit lesion with an antithrombotic agent is not required, nor is it efficient, as there is no benefit in terms of mortality or MI, and there is a higher rate of RI with such a strategy. Interestingly, a recent meta-analysis by Bellemain-Appaix et al. (26) failed to show any benefit of pre-treatment with P2Y12-ADP receptor antagonists prior to PCI. These findings were confirmed by the ACCOAST (A Comparison of Prasugrel at PCI or Time of Diagnosis of Non–ST-Elevation Myocardial Infarction) trial, in which pre-treatment with prasugrel was not associated with any ischemic benefit but a bleeding harm compared with using prasugrel at the time of PCI (27). Thus pre-treatment with an antithrombotic agent appears to have no potential benefit in preparing for a future intervention as proposed in a delayed invasive strategy.
In addition to this clinical benefit of the early strategy on the recurrence of ischemia, there is a potential health care benefit related to the reduction of the in-hospital LOS. This reduction in hospital LOS benefits patients and could translate into cost savings, although this was not evaluated in most of these RCTs (28).
Existing treatment guidelines are based largely on the meta-analysis by Navarese et al. (4) and the TIMACS trial. It is important to note that the TIMACS trial was terminated early and did not meet its primary endpoint (5). Therefore, its results should be interpreted with caution. In addition, the meta-analysis by Navarese et al. (4) did demonstrate trends in mortality and MI that were not confirmed in our updated analysis. The current guidelines classify patients according to their risk profiles, suggesting optimal delays for each subgroup (1); limited data exist regarding such subgroup classification. Only the TIMACS trial performed a pre-specified subgroup analysis on the basis of the GRACE (Global Registry of Acute Coronary Events) score. In this trial, patients with GRACEs score >140 had a significant decrease in the composite endpoint of death, MI, and stroke with an early invasive strategy (13.9% vs. 21%; p = 0.006) (4). This subgroup analysis of a trial with negative results should be considered hypothesis generating. Very high risk patients were excluded from most trials because they require immediate invasive management.
In the trials included in this meta-analysis, pre-treatment with dual-antiplatelet therapy was used. However, the use of a loading dose of a P2Y12-ADP receptor antagonist was recently challenged as described previously and is therefore controversial. Whether the lack of pre-treatment affects the optimal delay of the invasive strategy should be investigated in future RCTs.
The CIs around our estimates for mortality and MI are large and include values that could be clinically significant despite the addition of 3 trials including more than 1,000 patients to the previous meta-analysis.
The timing of the invasive strategy was variable among studies to define early and delayed strategies. In addition, the ischemic risk of the patients included varied among studies, with some excluding unstable angina and others including it.
Last, we did not explore the influence of PCI, coronary artery bypass grafting, and medical therapy in the sensitivity analyses, because the selected studies rarely described the outcomes according to these categories. Future trials should explore this important issue.
The present updated meta-analysis suggests that there is no difference in death or MI between early and delayed invasive strategies in patients with NSTE-ACS. However, an early strategy is safe and reduces both the rates of RI or RA and in-hospital LOS.
WHAT IS KNOWN? An invasive strategy is preferred in patients with NSTE-ACS to reduce death and MI.
WHAT IS NEW? There is no mortality difference between an early and a delayed invasive strategy in patients with NSTE-ACS. An early invasive strategy reduces RI and in-hospital LOS.
WHAT IS NEXT? This meta-analysis provides important data regarding the optimal timing of the invasive strategy in patients with NSTE-ACS. However, future trials should determine whether the results are different depending on subgroups of patients on the basis of their risk profiles. In addition, whether these results are valid for patients without pre-treatment with P2Y12 ADP receptor antagonists should be evaluated.
The authors thank Marie and Paul for the time devoted to this meta-analysis. They would also like to thank Emilie and Louise B. for their patience and careful work in data collection.
For supplemental figures and tables, please see the online version of this article.
Assistance-Publique Hôpitaux de Marseille provided technical support through grant PHRC-15-197. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- length of stay
- myocardial infarction
- non–ST-segment elevation acute coronary syndromes
- odds ratio
- percutaneous coronary intervention
- refractory angina
- randomized clinical trial
- recurrent ischemia
- Received June 29, 2016.
- Revision received August 29, 2016.
- Accepted September 8, 2016.
- American College of Cardiology Foundation
- ↵(2016) 2015 ESC guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation: Task Force for the Management of Acute Coronary Syndromes in Patients Presenting Without Persistent ST-Segment Elevation of the European Society of Cardiology (ESC). Eur Heart J 37:267–315.
- Riezebos R.K.,
- Ronner E.,
- Ter Bals E.,
- et al.,
- OPTIMA trial
- Thiele H.,
- Rach J.,
- Klein N.,
- et al.,
- for the LIPSIA-NSTEMI Trial Group
- Zhang J.,
- Qiao S.B.,
- Zhu J.,
- for the Chinese Cooperative Group of the Timing of Intervention in Acute Coronary Syndrome
- Milosevic A.,
- Vasiljevic-Pokrajcic Z.,
- Milasinovic D.,
- et al.
- van’t Hof A.W.,
- de Vries S.T.,
- Dambrink J.H.,
- et al.
- Reuter P.G.,
- Rouchy C.,
- Cattan S.,
- et al.
- Higgins J.P.,
- Thompson S.G.,
- Deeks J.J.,
- et al.
- Borenstein M.,
- Hedges L.V.,
- Higgins J.P.T.,
- et al.
- Borenstein M.,
- Hedges L.V.,
- Higgins J.P.T.,
- et al.
- Valkenhoef G, van Kuiper J. gemtc: network meta-analysis using Bayesian methods. Available at: http://cran.r-project.org/web/packages/gemtc/index.html. Accessed May 14, 2015.
- ↵(2012) Third universal definition of myocardial infarction. J Am Coll Cardiol 60:1581–1598.
- Bellemain-Appaix A.,
- Kerneis M.,
- O’Connor S.A.,
- et al.
- Dijksman L.M.,
- Hirsch A.,
- Windhausen F.,
- et al.,
- for the Invasive Versus Conservative Treatment in Unstable Coronary Syndromes Investigators