OUP user menu

Mortality in randomized controlled trials comparing drug-eluting vs. bare metal stents in coronary artery disease: a meta-analysis

Alain Joel Nordmann, Matthias Briel, Heiner Claudins Bucher
DOI: http://dx.doi.org/10.1093/eurheartj/ehl282 2784-2814 First published online: 4 October 2006

Abstract

Aims To evaluate the effect of drug-eluting vs. bare metal stents for the treatment of coronary artery disease on overall, cardiac, and non-cardiac mortalities.

Methods and results We conducted a systematic literature search to identify all randomized controlled trials comparing sirolimus or paclitaxel-eluting stents with bare metal stents and reporting mortality data after at least 1 year of follow-up. Trial data were reviewed and extracted independently by two investigators in an unblinded standardized manner. Seventeen trials including a total of 8221 patients were analysed. Peto's odds ratios (ORs) for total mortality after 1 (n=8221), 2 (n=4631), 3 (n=4105), and 4 (n=1293) years of follow-up were 0.94 [95% confidence interval (CI) 0.66–1.34], 1.11 (95% CI 0.76–1.61), 1.25 (95% CI 0.91–1.73), and 1.46 (95% CI 0.92–2.31), respectively. Corresponding ORs for non-cardiac mortality were 1.07 (95% CI 0.64–1.80), 1.72 (95% CI 1.01–2.94), 1.45 (95% CI 0.93–2.25), and 1.65 (95% CI 0.89–3.10). There was no difference in OR for cardiac mortality among all trials. In sensitivity analyses, sirolimus- but not paclitaxel-eluting stents were associated with an increase in non-cardiac mortality at 2 and 3 years of follow-up.

Conclusion Drug-eluting stents for the treatment of coronary artery disease do not reduce total mortality when compared with bare metal stents. Preliminary evidence suggests that sirolimus- but not paclitaxel-eluting stents may lead to increased non-cardiac mortality. Long-term follow-up and assessment of cause-specific deaths in patients receiving drug-eluting stents is mandatory to determine the long-term safety of these devices.

  • Drug-eluting stents
  • Meta-analysis
  • Coronary artery disease
See page 2737 for the editorial comment on this article (doi:10.1093/eurheartj/ehl378)

Introduction

In the past decade, stent implantation has become the treatment of choice among patients with coronary artery disease.1,2 Stents when compared with simple balloon angioplasty may reduce overall mortality in non-acute coronary artery disease3 and re-infarction rates in acute coronary artery disease.4 However, in-stent restenosis due to neointimal hyperplasia continues to limit the long-term success of coronary artery stenting.5,6 In order to resolve the problem of restenosis, drug-eluting stents were developed. In recently published meta-analyses of randomized controlled trials, drug-eluting stents were found to reduce restenoses and the need for revascularization procedures, but not to reduce overall mortality or the incidence of myocardial infarction.79 These meta-analyses were limited to a follow-up of 1 year after placement of the stents.

Despite some technical problems associated with drug-eluting stents10 and the publication of case reports of late thrombosis in drug-eluting stents after discontinuation of antiplatelet therapy,11 drug-eluting stents continue to be widely used. Recent estimates suggest that in the USA ≥90% of currently deployed coronary stents are drug-eluting stents.12 The vast majority of implanted drug-eluting stents is either sirolimus- or paclitaxel-eluting stents. A recent meta-analysis of head-to-head comparisons of these drug-eluting stents demonstrated a reduced need for revascularization procedures associated with sirolimus- when compared with paclitaxel-eluting stents.13

In this meta-analysis of randomized controlled trials, we evaluate the effects of drug-eluting stents compared with bare metal stents on overall, cardiac, and non-cardiac mortalities as well as the occurrence of stent thromboses during an extended follow-up for up to 4 years.

Methods

Literature search

We used the Cochrane Collaboration search strategy and searched MEDLINE, EMBASE, Web of Science, and the Cochrane Library from January 1980 up to April 2006 to identify all randomized controlled trials that compare sirolimus or paclitaxel-eluting stents with bare metal stents for the treatment of coronary artery disease. We additionally reviewed UptoDate version 2005 and Clinical Evidence Concise 2004 (issue 12), three websites dedicated to dissemination of results from cardiovascular trials (www.tctmd.com, www.theheart.org, www.clinicaltrialresults.org), contacted experts of the field, and searched reference lists of identified publications for citations of additional relevant articles. We contacted original trial investigators and stent manufacturers for additional information where needed.

Data processing and data extraction

Two investigators (A.J.N./M.B.) independently assessed trial eligibility and quality. Disagreement was resolved by consensus. Data of eligible trials were extracted in duplicate. We assessed the quality of included trials with respect to concealed treatment allocation, blinding of patients and caregivers, blinded outcome assessment, loss to follow-up, and full description of losses to follow-up and withdrawals.14

Eligible trials were required to have a randomized controlled design comparing drug-eluting sirolimus or paclitaxel stents with bare metal stents in patients with coronary artery disease, to report mortality data, and to have a follow-up of at least 1 year. Trials were excluded if they compared drug-eluting with bare metal stents in non-native coronary arteries, if they used other eluting drugs than sirolimus or paclitaxel (e.g. everolimus), and if they compared drug-eluting stents with each other.

Data aggregation

We calculated Peto's odds ratios (ORs) for total, cardiac, and non-cardiac mortalities and stent thromboses based on intention-to-treat data. Simulation suggests that with rare events, meta-analysis underestimates the true effect of treatment over control, but Peto's ORs are the least biased and most powerful method of pooling trial results among the methods in common use.15,16 When events are rare, ORs are a close approximation to the relative risk.17

We tested for heterogeneity with the Cochran Q-test and measured inconsistency (I2; the percentage of total variance across studies that is due to heterogeneity rather than chance) of treatment effects across trials.18 In addition, we performed sensitivity analyses for all outcomes comparing trials using sirolimus vs. trials using paclitaxel-eluting stents.

We investigated the presence of publication bias by means of funnel plots.19 All statistical analyses were performed using Stata 8.2 (StataCorp, College Station, TX, USA).

Results

We identified 17 trials including a total of 8221 patients which reported mortality data after 1 year. Seven trials (n=2487) used sirolimus, nine (n=4908) used paclitaxel, and one20 (n=826) used both sirolimus- and paclitaxel-eluting stents. Figure 1 illustrates the flow of the selection process for potentially eligible trials and reasons for exclusion. Funnel plots on total, cardiac, and non-cardiac mortalities from year 1 to 3 indicated little evidence for the presence of publication bias (Egger's test consistently P≥0.1). Twelve trials including 4631 patients reported mortality data after 2 years. Nine trials including 4105 patients reported mortality data after 3 years, and two trials including 1293 patients reported mortality data after 4 years of follow-up.

Figure 1

Flow chart on selection of included trials. DES, drug-eluting stent.

Most trials were designed to assess the efficacy of drug-eluting stents at 9–12 months after the index percutaneous coronary intervention in de novo (not restenotic) lesions. Thus, full-paper publications mostly reported 9-month or 1-year follow-up results. Later, follow-up information was retrieved from either web-posted conference proceedings, personal communication with the principle investigators, or stent manufacturers. The RAVEL,21 SIRIUS,2224 and TAXUS2428 trials pre-specified reporting of long-term follow-up for up to 5 years, whereas in all other trials protocol-defined follow-up ended after 6 months,29 9 months,3012 months,31 18 months,20 and 2 years.3234 Requested collaboration was obtained in all cases.

Mean age of patients enrolled in individual trials ranged between 56 and 67 years and patients were mostly male (Table 1). Only one trial included patients with acute ST-elevation myocardial infarctions (21% of patients included in the BASKET-trial20). One trial included only patients with diabetes.33 Antiplatelet therapy with clopidogrel, ticlopidine, or cilostazol was recommended in all trials for at least 2–6 months after percutaneous coronary intervention. One publication included results of two trials; one comparing the TAXUS MR stent wirh a control group and the other comparing the TAXUS SR stent to a different control group.26

View this table:
Table 1

Characteristics of included trials

AuthorTrialsType of drugNumber of patients in DES/BMSFollow-up (months)Mean (SD) age (years)Men (%)Diabetes (%)Prior MI (%)Reference diameter (mm)Lesion Length (mm)Post-interventional prophylaxisConcealed treatment allocationBlinding of patients/caregivers/assessors
Morice et al.21,38,39RAVELSirolimus120/11848627016362.69.6Aspirin indefinitely. Clopidogrel or ticlopidine for 2 monthsYesYes/yes/yes
Moses et al. 22,40,41,SIRIUSSirolimus533/52548627325312.814.4Aspirin indefinitely. Clopidogrel for 3 monthsYesYes/yes/yes
Schofer et al.23,42E-SIRIUSSirolimus175/17736627019422.615.0Aspirin indefinitely. Clopidogrel or ticlopidine for 2 monthsYesYes/yes/yes
Grube et al.25,43TAXUS IPaclitaxel31/303666891828311.3Aspirin for at least 1 year. Clopidogrel for 6 monthsUnclearYes/yes/yes
Colombo et al. MR26TAXUS II MRPaclitaxel135/13436595612302.810.5Aspirin indefinitely. Clopidogrel or ticlopidine for at least 6 monthsUnclearYes/yes/yes
Colombo et al. SR26TAXUS II SRPaclitaxel131/13636615610292.810.5Aspirin indefinitely. Clopidogrel or ticlopidine for at least 6 monthsUnclearYes/yes/yes
Park et al.35ASPECTPaclitaxel117/5924588018243.211Aspirin indefinitely. Clopidogrel, ticlopidine or cilostazol for 1 to 6 months.YesYes/yes/yes
Schampaert et al.24,44C-SIRIUSSirolimus50/5036607024452.713.6Aspirin indefinitely. Clopidogrel for 2 monthsYesYes/yes/yes
Lansky et al.30DELIVERPaclitaxel517/51212627131262.911.3Aspirin for 1 year. Clopidogrel for 3 monthsUnclearNot stated in publication
Gershlick et al.32ELUTESPaclitaxel152/382456811135310.8Aspirin and clopidogrel for 3 monthsYesYes/yes/yes
Ardissino et al.34SES-SMARTSirolimus129/12812637225282.211.8Aspirin and clopidogrel for at least 2 monthsYesYes/yes/yes
Stone et al.27,45TAXUS IVPaclitaxel662/65236637223312.813.4Aspirin indefinitely. Clopidogrel for 6 monthsYesYes/yes/yes
Dawkins et al.28,46TAXUS VIPaclitaxel219/22736637620422.820.6Aspirin indefinitely. Clopidogrel for 6 monthsYesYes/yes/yes
Stone et al.47TAXUS VPaclitaxel577/57912637031282.717.3Aspirin indefinitely. Clopidogrel for at least 6 monthsYesYes/yes/yes
Sabate et al.33DIABE-TESSirolimus80/80246763100372.415.0Aspirin indefinitely. Clopidogrel for 1 year.YesNo/No/yes
Kelbaek31SCAND-STENTSirolimus163/15912637618522.918.0Aspirin indefinitely. Clopidogrel for 1 year.YesNo/no/yes
Kaiser20 et al.BASKETSirolimus2641264791927NANAAspirin indefinitely. Clopidogrel for 6 monthsUnclearNo/no/yes
Paclitaxel281
281 (BMS)

Two trials compared the effects of bare metal stents with various doses of paclitaxel-eluting stents.25,35 For the purpose of this analysis, we compared patients assigned to bare metal stents with all patients assigned to any dose of paclitaxel-eluting stents.

Event rates for total, cardiac, and non-cardiac mortalities in individual trials comparing sirolimus and paclitaxel-eluting stents are summarized in Table 2.

View this table:
Table 2

Data sources and event rates for overall, cardiac, and non-cardiac mortalities in randomized controlled trials comparing sirolimus- and paclitaxel-eluting stents to bare metal stents

TrialData sourceFollow-up (years)Event rates in patients treated with DES and BMS (%)
DeathCardiac deathNon-cardiac death
RAVEL (2002)P1Sirolimus1.701.7
BMS1.70.850.85
CP/PI2Sirolimus505
BMS2.51.70.8
CP/PI3Sirolimus7.50.86.7
BMS4.22.51.7
CP4Sirolimus10.82.58.3
BMS5.94.21.7
SIRIUS (2003)CP/PI1Sirolimus1.30.60.7
BMS0.80.40.4
P2Sirolimus2.10.81.1
BMS1.30.60.4
M3Sirolimus3.92.11.9
BMS2.91.71.1
M4Sirolimus63.22.8
BMS4.62.12.4
E-SIRIUS (2003)P1Sirolimus1.10.60.6
PBMS0.600.6
PI2Sirolimus2.31.11.1
BMS2.82.30.6
M3Sirolimus5.141.1
BMS3.92.81.1
TAXUS I (2003)P1Paclitaxel000
BMS000
M2Paclitaxel3.203.2
BMS000
M3Paclitaxel3.203.2
BMS000
TAXUS II MR (2003)P1Paclitaxel0.700.7
BMS0.700.7
CP/M2Paclitaxel30.72.3
BMS1.501.5
CP/M3Paclitaxel3.71.52.2
BMS2.202.2
TAXUS II SR (2003)P1Paclitaxel000
BMS1.51.50
M2Paclitaxel1.60.80.8
BMS2.22.20
M3Paclitaxel4.61.53.1
BMS2.92.20.7
ASPECT (2003)P/PI1Paclitaxel0.90.90
BMS000
PI2Paclitaxel0.90.90
BMS000
C-SIRIUS (2004)CP/PI1Sirolimus000
BMS000
CP/PI2Sirolimus202
BMS000
M3Sirolimus422
BMS000
DELIVER (2004)M1Paclitaxel10.20.8
BMS1.40.80.6
ELUTES (2004)P1Paclitaxel0.70.70
BMS000
PI2Paclitaxel0.70.70
BMS000
SES-SMART (2004)PI1Sirolimus0.800.8
BMS3.90.83.1
TAXUS IV (2004)P1Paclitaxel2.11.40.8
BMS1.71.20.5
M2Paclitaxel3.61.81.8
BMS2.62.11.5
M3Paclitaxel3.91.42.6
BMS4.11.52.6
TAXUS V (2005)P/CP/M1Paclitaxel2.111
BMS1.710.7
TAXUS VI (2005)P/M1Paclitaxel000
BMS1.80.90.9
CP2Paclitaxel1.41.40
BMS1.40.50.9
M3Paclitaxel2.31.40.9
BMS3.22.30.9
DIABETES (2004)P/PI1Sirolimus3.81.32.5
BMS52.52.5
PI2Sirolimus6.32.53.8
BMS6.33.82.5
SCANDSTENT (2005)CP/PI1Sirolimus0.60.60
BMS0.60.60
BASKET (2005)PI1Sirolimus2.71.51.1
Paclitaxel3.62.80.7
BMS3.62.11.4
  • BMS, bare metal stents; DES, drug-eluting stents; NA, not available; P, publication; CP, conference proceeding; PI, principal investigator; M, manufacturer.

Quality of trials

In general, the quality of included trials was very good. All trials reported intention-to-treat analyses. All but three trials20,26,30 reported concealed treatment allocation. All but four trials20,30,31,33 reported blinding of patients and caregivers. In all trials, clinical endpoints were adjudicated by an independent clinical events committee (blinded outcome assessment). Four trials did not fully describe the reasons for withdrawals and losses to follow-up.21,26,27,31 All trials had a loss of follow-up of <10% at the time of publication of the main pre-specified outcome report (6–12-month follow-up examinations). No trial reported rates of patients lost to follow up after 1 year.

Overall mortality

ORs for overall mortality in patients treated with drug-eluting when compared WITH patients treated with bare metal stents were after 1 year (n=17 trials) 0.94 [95% confidence interval (CI) 0.66–1.34; test for heterogeneity P=0.6], after 2 years (n=12 trials) 1.11 (95% CI 0.76–1.61; test for heterogeneity P=0.7), after 3 years (n=9 trials) 1.25 (95% CI 0.91–1.73; test for heterogeneity P=0.9), and after 4 years (n=2 trials) 1.46 (95% CI 0.92–2.31; test for heterogeneity P=0.5) (Figure 2).

Figure 2

Forest plots on overall, cardiac and non-cardiac mortalities in randomized controlled trials comparing drug-eluting stents with bare metal stents in coronary artery disease. BMS, bare metal stent; DES, drug-eluting stent; UI, uncertainty interval. (Asterisk) The BASKET trial had three arms (a bare metal stent, a sirolimus- and a paclitaxel-eluting stent arm). Thus, there are 17 trials in total, eight comparisons between sirolimus-eluting and bare metal stents, and 10 comparisons between paclitaxel-eluting and bare metal stents.

Cardiac mortality

ORs for cardiac mortality in patients treated with drug-eluting when compared with patients treated with bare metal stents were after 1 year (n=17 trials) 0.84 (95% CI 0.52–1.36; test for heterogeneity P=0.7), after 2 years (n=12 trials) 0.73 (95% CI 0.44–1.23; test for heterogeneity P=0.8), after 3 years (n=9 trials) 1.00 (95% CI 0.62–1.60; test for heterogeneity P=0.6), and after 4 years (n=2 trials) 1.24 (95% CI, 0.64–2.40; test for heterogeneity P=0.2).

Non-cardiac mortality

The OR for non-cardiac mortality in patients treated with drug-eluting compared with patients treated with bare metal stents was not different after 1 year of follow-up (n=17 trials) (OR 1.07; 95% CI 0.64–1.80; test for heterogeneity P=0.8). However, from 2 to 4 years of follow-up, OR for non-cardiac mortality tended to be higher in patients treated with drug-eluting than in patients treated with bare metal stents: After 2 years (n=12 trials), the OR was 1.72 (95% CI 1.01–2.94; test for heterogeneity P=0.6), after 3 years (n=9 trials) 1.45 (95% CI 0.93–2.25; test for heterogeneity P=0.7), and after 4 years (n=2 trials) 1.65 (95% CI 0.88–3.10; test for heterogeneity P=0.08).

Sensitivity analyses

Sensitivity analyses did not reveal any difference in overall or cardiac mortality between trials using sirolimus- and trials using paclitaxel-eluting stents. However, at 2 and 3 years of follow-up, non-cardiac mortality was significantly increased when compared with bare metal stents in trials using sirolimus-, but not paclitaxel-eluting stents (OR 2.74, 95% CI 1.22–6.13; test for heterogeneity P=0.5 in trials using sirolimus-eluting stents compared with OR 1.21, 95% CI 0.59–2.45; test for heterogeneity P=0.4 in trials using paclitaxel-eluting stents at 2 years, and OR 2.04, 95% CI 1.00–4.15; test for heterogeneity P=0.6 in trials using sirolimus-eluting stents compared with OR 1.17, 95% CI 0.67–2.05; test for heterogeneity P=0.6 in trials using paclitaxel-eluting stents at 3 years) (Figure 2). Cause-specific reasons for non-cardiac deaths in patients treated with sirolimus-eluting stents are summarized in Table 3.

View this table:
Table 3

Non-cardiac deaths in patients randomized to sirolimus-eluting stents

TrialFollow-up (years)nCause of death
RAVEL4104×cancer (lung, prostate, pancreas, gastro-intestinal)
3×stroke (2×haemorrhage, 1×ischaemia)
2×pneumonia
1×pulmonary embolism
SIRIUS4154×cancer (2×renal, 1×colon, 1×unspecified)
4×respiratory failure
2×stroke (1×haemorrhage, 1×ischaemia)
2×sepsis
1×lymphoma
1×car accident
1×seizure disorder (Alzheimer's disease)
C-SIRIUS311×cancer (lung)
E-SIRIUS321×cancer (lung)
1×sepsis
SES-SMART111×pancreatic cancer
DIABETES131×septic shock
1×stroke
1×pulmonary embolism
BASKET133×cancer (2×colon cancer, 1×rectum cancer)*
  • *2 cancers already known at time of trial inclusion.

  • No non-cardiac deaths in patients randomized to drug-eluting stents so far in the SCANDSTENT trial.

Stent thromboses

In general, stent thromboses occurring 30 days or more after stent implantation were rare (Table 4). There was no difference in the incidence of early or late stent thromboses in patients treated with drug-eluting stents compared with patients treated with bare metal stents (OR 1.15, 95% CI 0.63–2.11; test for heterogeneity P=0.7 for stent thromboses occurring ≤30 days after stent implantation, and OR 1.33, 95%CI 0.68–2.58; test for heterogeneity P=0.13 for stent thromboses occurring from 30 days to the latest available follow-up after stent implantation).

View this table:
Table 4

Stent thrombosis in trials of sirolimus or paclitaxel-eluting stents vs. bare metal stents in coronary artery disease

Trial (reference)Eluting drugRandomized individuals (n)Follow-up (months)Subacute (≤ 30 days), n (%)31–365 days, n (%)1–2 years, n (%)2–3 years, n (%)3–4 years, n (%)
DESBMSDESBMSDESDESBMSBMSDESBMSDESBMS
RAVEL (2002)Sirolimus120118480000000000
SIRIUS (2003)Sirolimus533525481 (0.2)1 (0.2)1 (0.2)3 (0.6)1 (0.2)01 (0.2)01 (0.2)0
E-SIRIUS (2003)Sirolimus175177362 (1.1)0001 (0.6)01 (0.6)0
C-SIRIUS (2004)Sirolimus5050361 (2.0)001 (2.0)0000
TAXUS I (2003)Paclitaxel313036a0000000000
TAXUS II (MR+SR) (2003)Paclitaxel266270361 (0.4)02 (0.8)03 (1.1)001 (0.4)
ASPECT (2003)Paclitaxel11759244 (3.4)00000
DELIVER (2004)Paclitaxel517512121 (0.2)1 (0.2)1 (0.2)1 (0.2)
ELUTES (2004)Paclitaxel15238241 (0.7)1 (2.6)0000
SES-SMART (2004)Sirolimus12912812b1 (0.8)1 (0.8)03 (2.3)
TAXUS IV (2004)Paclitaxel662652362 (0.3)4 (0.6)2 (0.3)1 (0.2)3 (0.5)000
TAXUS VI (2005)Paclitaxel219227361 (0.5)2 (0.9)001 (0.5)000
TAXUS V (2005)Paclitaxel577579124 (0.7)3 (0.5)01 (0.2)
DIABETES (2005)Sirolimus8080240001 (1.3)2 (2.5)1 (1.3)
SCANDSTENT (2005)Sirolimus16315912c1 (0.6)3 (1.9)02 (1.3)
BASKET (2005)Sirolimus264281123 (1.1)2 (0.7)
BASKET (2005)Paclitaxel2813 (1.1)
  • BMS, bare metal stents; DES, drug-eluting stents.

  • aMortality data available up to 36 months of follow-up, data on stent thromboses available for up to 48 months of follow-up.

  • bData on stent thromboses available for 8 months follow-up only.

  • cData on stent thromboses available for 9 months follow-up only.

Discussion

In this meta-analysis of randomized controlled trials, we found a trend towards an increased risk for overall mortality in patients treated with drug-eluting stents compared with bare metal stents among trials providing data from the second to the fourth year of follow-up. However, CIs for this finding were wide and included 1. Although there was no difference in cardiac mortality, non-cardiac mortality seemed to be higher among patients treated with drug-eluting than among patients treated with bare metal stents. In sensitivity analyses comparing trials using sirolimus- and trials using paclitaxel-eluting stents, this increase in non-cardiac mortality seemed to be limited to trials using sirolimus-eluting stents.

We have conducted an extensive literature search to retrieve all relevant eligible trials. Although formal testing did not indicate any publication bias, such a bias cannot definitely be ruled out due to the small number of trials included and the low power of any test to detect a publication bias. Although publication of the TAXUS trials2528 generally did not present data on overall, but only on cardiac mortality, we were able to obtain data on overall mortality for all trials by directly contacting the stent manufacturer. The quality of the included trials was generally very good, and there was no evidence of heterogeneity among treatment effects on mortality.

Our study has several limitations and should be interpreted with caution. Most non-cardiac deaths in patients treated with sirolimus-eluting stents were either due to cancer, stroke, or infectious diseases. Owing to the small number of non-cardiac deaths in the included trials, we were not able to explore whether one of these individual cause-specific deaths occurred more commonly in patients treated with sirolimus-eluting stents than in patients treated with bare metal stents. Therefore, we cannot rule out that the observed association of an increase in non-cardiac mortality with the use of sirolimus-eluting stents might be due to chance alone. However, especially the association of cancer-related deaths associated with the use of sirolimus-eluting stents requires further attention. There is little evidence in the medical literature about serious side effects associated with the use of sirolimus. However, a large multicentre trial to evaluate pre-emptive sirolimus therapy in recipients of orthotopic liver transplantation reported an increased incidence of death in the group treated with sirolimus and tacrolimus when compared with the group treated with tacrolimus alone.36 A further limitation of our analysis is the lack of information on differential loss of follow-up after regular termination of trials. Follow-up data were in majority retrieved from web-posted conference proceedings or personal correspondence with either the principle investigators of trials or stent manufacturers. All trials including sirolimus-eluting stents pre-specified the reporting of total mortality. In contrast, publications of the TAXUS trials did not routinely report overall mortality, but were restricted to the report of cardiac mortality as an individual component of a combined primary outcome of major cardiovascular events. The database of all TAXUS trials is entirely under control of the stent manufacturer (Boston Scientific Corporation, USA). We were therefore unable to verify the reported mortality rates in the TAXUS trials and had to rely on information provided by the manufacturer. Non-cardiac mortality was not pre-specified in any trial. Any potential bias originating from a biased data source such as the stent manufacturer, however, would introduce a bias towards a lack of association between the use of drug-eluting stents and increased mortality. Therefore, the findings of our analysis should all the more so raise concern about the safety of drug-eluting stents. This concern is supported by the results of the late BASKET trial that observed an increase in a combined outcome of cardiac death and non-fatal myocardial infarction after 18 months in patients treated with drug-eluting stents when compared with patients treated with bare metal stents.37 The preliminary findings from this meta-analysis demand full disclosure of cause-specific mortality data in all trials with drug-eluting stents with extended observation periods. The databases of randomized controlled trials comparing drug-eluted stents with bare metal stents should be controlled by an independent organization and not by the stent manufacturers. Our results further stress the importance of monitoring long-term safety of drug-eluted devices in cohort studies.

In summary, our results fail to demonstrate a mortality benefit for drug-eluting stents when compared with bare metal stents for the treatment of coronary artery disease during an extended follow-up of up to 4 years. In the absence of any difference in cardiac mortality, the use of sirolimus-eluting stents seems to be associated with an increase in non-cardiac mortality. Long-term surveillance of patients treated with drug-eluted stents and full disclosure of causes of deaths of these patients are crucial for the assessment of the long-term safety of drug-eluted stents.

Acknowledgements

We thank Santésuisse and the Gottfried Bangerter-Rhyner-Foundation for supporting our work. We thank all principle investigators of included trials for providing us with original data from their trials. We also thank Joerg Koglin from Boston Scientific Corporation, Boston, USA, for providing us with data from the TAXUS trials. We thank Gordon Guyatt, Department of Biostatistics and Clinical Epidemiology, McMaster University, Hamilton, Canada, for critically reviewing the manuscript. All the authors are supported by Santésuisse and the Gottfried and Julia Bangerter-Rhyner-Foundation. The funding source had no role in the study design, in the collection, analysis, and interpretation of data, in the writing of the report, and in the decision to submit the paper for publication. All the listed authors have substantially contributed to the conception and design of the study, were involved in the analysis and interpretation of data, drafted the article or revised it critically for important intellectual content, and approved the final version of the manuscript.

Conflict of interest: none declared.

Appendix

Funnel plots

Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image Embedded Image

Forest plots

References

View Abstract