Early Invasive Versus Selectively Invasive Strategy inPatients With Non-ST-Segment Elevation Acute
Coronary Syndrome: Impact of Age
Fabio Angeli,1* MD, Paolo Verdecchia,2 MD, Stefano Savonitto,3 MD, Nuccia Morici,4 MD,Stefano De Servi,5 MD, and Claudio Cavallini,6 MD
Background: It is unclear whether the benefits of an early invasive strategy (EIS) inpatients with non-ST-segment elevation acute coronary syndromes (NSTEACS) equallyapply to younger and older individuals. Elderly patients are generally less likely toundergo EIS when compared with younger patients. Objectives: We conducted ameta-analysis to compare the benefit of an EIS versus a selectively invasive strategy(SIS) in patients with NSTEACS. We tested the hypothesis that the magnitude of benefitof an EIS over a SIS mainly applies to older individuals. Methods: We extracted datafrom randomized controlled trials (RCTs) identified through search methodology filters.The primary outcome of the analysis was the composite of all-cause death and myo-cardial infarction (MI). Secondary outcomes were death and MI taken alone and re-hospitalization. Results: Nine trials (n 5 9,400 patients) were eligible. The incidence ofthe composite end-point of MI and all-cause death was 16.0% with the EIS and 18.3%with the SIS (OR: 0.85, 95% CI: 0.760.95). The incidence of MI was 8.4% with the EISand 10.9% with the SIS (OR: 0.75, 95% CI: 0.660.87). Similar results were obtained forrehospitalization (OR: 0.71, 95% CI: 0.550.90). The incidence of all-cause death did notdiffer between the two groups. The EIS reduced the composite end-point andre-hospitalization to a greater extent in elderly than in younger patients (P for inter-action 5 0.044 and
revascularization during the early phase of index hospi-talization, some findings from clinical studies suggest alack of significant benefit of an early invasive strategy(EIS) in low-risk patients . For this reason, currentguidelines  recommend an EIS over a more con-servative strategy for the management of patients withNSTEACS at high risk for recurrent events.
Older age itself is a major risk factor for adverseoutcome in NSTEACS and reports from individual tri-als showed that the benefit from an invasive strategy ismainly observed in patients >65 years of age [10,11].Nevertheless, in real-world practice NSTEACS elderlypatients are substantially less likely to undergo an inva-sive strategy than younger patients and many centerstend to use a routine EIS (including the routine use ofdiagnostic catheterization and revascularization as indi-cated) in younger patients, and a more selectiveapproach (i.e., selectively invasive strategy [SIS]) inolder individuals . These data emphasize thegrowing importance of developing invasive strategymodalities in elderly patients with NSTEACS .
Thus, we performed a meta-analysis to further eluci-date the benefit of an EIS compared to a more con-servative management of NSTEACS patients. Inparticular, we explored outcome in relation to age anddifferent patient- and trial-level characteristics in orderto identify subsets of patients with NSTEACS whowould benefit most from an early invasive approach.
We extracted data from randomized controlled trials(RCTs) evaluating the prognostic implications of anEIS in NSTEACS patients and met all of the followingselection criteria:
i. diagnosis of NSTEACS as inclusion criteria forRCTs;
ii. allocation of at least 100 patients to an EIS or aninitially conservative management, where early inter-vention was defined as coronary angiography per-formed within 4 days from hospitalization orrandomization. Initially conservative managementwas defined by early conventional medical therapyor selective invasive management if indicated by re-fractory or recurrent symptoms or inducible ischemia(ischemia-driven or symptom-driven angiography);
iii. outcome data ascertained during a follow-up of atleast 6 months;
iv. publication before 30 April 2013 in peer-reviewedjournals indexed in Medline and Google Scholar. Wealso searched clinical trial registries (www.clinicaltrials.gov and www.controlled-trials.com) for unpublishedstudies.
We searched for eligible studies, using researchMethodology Filters . The following research termswere used: randomized controlled trials, myocardialinfarction, acute coronary syndrome, angioplasty,balloon, coronary, percutaneous coronary inter-vention, revascularization, coronary artery bypassgraft, outcome, and prognosis. We checked thereference list of identified articles and previous system-atic reviews to find other potentially eligible studies.We also performed hand-searching of conference pro-ceedings, pharmaceutical industry files and personalcommunication from experts in the field , to iden-tify any other relevant study. No language restrictionwas applied, to avoid discriminating papers not writtenin the English language (tower of Babel bias) .
Data extraction was done independently by twoauthors (PV and FA) and divergences were discussedand resolved in conference. The present meta-analysiswas done according to established methods and to theQuality of Reporting of Meta-analyses (QUORUM)statement [22,23].
Outcome of patients allocated to an EIS was com-pared to NSTEACS patients randomized to an initiallyconservative management. The primary outcome of theanalysis was the composite of all-cause death and MI.Secondary outcomes were death and MI taken aloneand rehospitalization.
We accepted the definition of NSTEACS as reportedin the individual clinical trials. Recurrent MI was gen-erally defined as new or recurrent ischemic symptomsand an increase of the creatine kinase-MB (CK-MB) ortroponin levels or electrocardiographic changes. Reho-spitalization was generally defined as readmission forcardiovascular causes.
For each end point, the longest follow-up availablein each study was chosen.
We calculated odds ratios (ORs) and 95% confi-dence intervals (CIs) for all-cause death, myocardialreinfarction, rehospitalization and for the compositeend-point of overall mortality and myocardial reinfarc-tion for each clinical study separately and for the com-bination of studies according to random- and fixed-effects models [23,24]. The null hypothesis of homoge-neity across individual studies was tested by using theQ-test. Pooled estimates were assessed for heterogene-ity by using the I2 and s2 statistics . Typically, val-ues of I2 above 50% are deemed to suggest largeamong-study heterogeneity, values of 2550% aredeemed to show moderate heterogeneity, and valuesbelow 25% denote low heterogeneity . Heterogene-ity was explored by conducting the analysis in the fol-lowing sets of pre-planned subgroups  of RCTs: (i)mean age of patients (
invasive strategy arm (24 hr; >24 hr); (iii) publica-tion year (before 2001 vs. after 2001); (iv) prevalenceof diabetes mellitus (12 months).
The potential effect modifiers, as identified fromsubgroup analysis, were further evaluated by randomeffect meta-regressions . This method uses eachstudys estimate and its standard error to separatewithin-study from between-study variability .
The fit of each model was assessed using the per-centage of among-study variance explained, togetherwith a significance test for each covariate. Reductionof among-study variance was estimated by the changein t2 statistics .
Publication bias was tested by visual inspection ofthe funnel plot and, more formally, with a modifiedregression test based on sample size .
Analyses were performed using Stata, version 12(StataCorp LP, College Station, Texas) and R version2.9.2 (R Foundation for Statistical Computing, Vienna,Austria).
Overall, we screened 1586 records. We identified 32relevant clinical trials (Fig. 1). Of these 32 trials[11,16,17,2962], 12 studies were excluded for beingnoncontemporary  (i.e., low use of dual oral anti-platelet therapy and stenting, or use of fibrinolyticagents), 8 studies were excluded because they had com-pared an early to a delayed invasive strategy  orearly systematic angiography and revascularisation versusearly elective strategy  (angiography and selectiverevascularisation), 1 was excluded because it hadrandomized only 88 patients , 1 was excluded beinga subanalysis on female patients  and 1 trial wasexcluded because it was a randomized comparison ofcoronary-artery bypass surgery and medical therapy .
Nine trials [11,16,17,2939] involving 9,400 patients(4702 EIS and 4698 more conservative management)met the inclusion criteria and were included in theanalysis.
Figure 1 shows the flow diagram with informationabout the selected, included, and excluded clinical studies.
Main Features of RCTs
Table I summarises the main features of trialsincluded in the analysis. Of the nine trials[11,16,17,2939], five reported long-term follow-up[11,29,30,3237]. Three RCTs included patients with a
mean age >65 years [11,17,29,36,39]. Prevalence ofdiabetes mellitus and history of MI ranged from 12[11,29,36] to 36% [16,17] and from 22 [11,29,36] to43% , respectively.
Individual and pooled ORs for the composite end-point of mortality and myocardial reinfarction in theRCTs are depicted in Fig. 2. The EIS was associatedwith a significant reduction in the risk of the compositeend-point when compared to an initially conservativemanagement (OR 0.85, 95% CI: 0.760.95, P 0.004).However, there was significant heterogeneity acrossthe studies (I2 50.4%). The extent of benefit waslarger in RCTs with a mean age of 65 years (OR0.75; 95% CI: 0.630.91) as compared to trials withyounger study populations (P 0.044 for heterogeneitybetw