Drug‐eluting stents versus bare‐metal stents for acute coronary syndrome

Joshua Feinberg; Emil Eik Nielsen; Janette Greenhalgh; Juliet Hounsome; Naqash J Sethi; Sanam Safi; Christian Gluud; Janus C Jakobsen

doi:10.1002/14651858.CD012481.pub2

استنت‌‌ها با پوشش دارویی در مقابل استنت‌‌های فلزی بدون پوشش در مدیریت سندرم کرونری حاد

Declaraciones de intereses de los autores

Versión publicada: 23 agosto 2017 Historial de versiones

https://doi.org/10.1002/14651858.CD012481.pub2

Contraer todo Desplegar todo

چکیده

disponible en

پیشینه

در سال 2012 تقریبا 3.7 میلیون نفر در سراسر جهان در اثر ابتلا به سندرم کرونری حاد فوت کردند. سندرم کرونری حاد، که همچنین تحت نام انفارکتوس میوکارد یا آنژین پکتوریس ناپایدار شناخته شده، به دلیل انسداد ناگهانی عروق خونی تامین کننده ماهیچه قلب ایجاد می‌شود. مداخله کرونری از راه پوست (percutaneous coronary intervention) اغلب برای سندرم کرونری حاد استفاده می‌شود، اما مطالعات مروری سیستماتیک قبلی در خصوص مقایسه اثرات استنت‌های دارای پوشش دارویی (drug‐eluting stents) با استنت‌های فلزی بدون پوشش (bare‐metal stents) نتایج متناقضی از نظر انفارکتوس میوکارد نشان داده‌اند؛ به‌طور کامل خطر خطاهای تصادفی و سیستماتیک را در نظر نگرفته؛ و شامل تمام کارآزمایی‌های بالینی تصادفی‌سازی شده مرتبط نبوده‌اند.

اهداف

ارزیابی مزایا و مضرات استنت‌های دارای پوشش دارویی در مقابل استنت‌های فلزی بدون پوشش در افراد مبتلا به سندرم کرونری حاد.

روش‌های جست‌وجو

پایگاه ثبت مرکزی کارآزمایی‌های کنترل شده کاکرین (CENTRAL)؛ MEDLINE؛ Embase؛ منابع علمی سلامت آمریکای لاتین و کارائیب (LILACS)؛ SCI‐EXPANDED، و BIOSIS را از آغاز به کار تا ژانویه 2017 جست‌وجو کردیم. همچنین دو پایگاه ثبت کارآزمایی‌های بالینی، بانک‌های اطلاعاتی آژانس دارویی اروپا (European Medicines Agency) و اداره غذا و داروی ایالات متحده (US Food and Drug Administration) و وب‌سایت‌های شرکت دارویی را بررسی کردیم. به‌ علاوه، فهرست منابع مقالات مطالعه مروری و کارآزمایی‌های مرتبط را جست‌وجو کردیم.

معیارهای انتخاب

کارآزمایی‌های بالینی تصادفی‌سازی شده که اثرات استنت‌های دارویی را در مقابل استنت‌های فلزی بدون پوشش برای سندرم کرونری حاد ارزیابی کردند. کارآزمایی‌ها را بدون توجه به نوع انتشار، وضعیت، تاریخ، یا زبان آنها وارد کردیم.

گردآوری و تجزیه‌وتحلیل داده‌ها

پروتکل منتشر شده و پیشنهادهای روش‌شناسی کاکرین خود را دنبال کردیم. دو نویسنده مرور به‌طور مستقل از هم داده‌ها را استخراج کردند. خطرات خطای سیستماتیک را بر اساس حوزه‌های سوگیری (bias) ارزیابی کردیم. تجزیه‌وتحلیل‌های مرحله‌‏ای کارآزمایی (Trial Sequential Analyses) را برای کنترل خطرات خطاهای تصادفی انجام دادیم. پیامدهای اولیه ما مورتالیتی به هر علتی، وقایع مهم قلبی‌عروقی، عوارض جانبی جدی، و کیفیت زندگی بودند. پیامدهای ثانویه عبارت بودند از آنژین، مورتالیتی قلبی‌عروقی، و انفارکتوس میوکارد. نقطه زمانی ارزیابی اولیه، در حداکثر دوره پیگیری بود. کیفیت شواهد را با رویکرد درجه‌‏بندی توصیه‏، ارزیابی، توسعه و ارزشیابی (GRADE) ارزیابی کردیم.

نتایج اصلی

ما 25 کارآزمایی را انتخاب کردیم که مجموعا 12,503 شرکت‌کننده را تصادفی‌سازی کردند. همه کارآزمایی‌ها در معرض خطر بالای سوگیری بوده، و سطح کیفیت شواهد طبق GRADE از پائین تا بسیار پائین متغیر بود. ما 22 کارآزمایی را وارد کردیم که در آن شرکت‌کنندگان مبتلا به انفارکتوس میوکارد با ST افزایش‌ یافته بودند، 1 کارآزمایی که در آن شرکت‌کنندگان مبتلا به انفارکتوس میوکارد بدون ST افزایش‌ یافته بودند، و 2 کارآزمایی که در آنها شرکت‌کنندگان ترکیبی را از سندرم‌های کرونری حاد داشتند.

متاآنالیزها در حداکثر دوره پیگیری نشان داد که هیچ تفاوتی در مقایسه میان استنت‌های دارای پوشش دارویی و استنت‌های فلزی بدون پوشش از نظر خطر مورتالیتی به هر علتی یا وقایع مهم قلبی‌عروقی وجود نداشت. خطر مطلق مرگ 6.97% در گروه استنت‌های دارای پوشش دارویی در مقایسه با 7.74% در گروه استنت‌های فلزی بدون پوشش (خطر نسبی (RR): 0.90؛ 95% فاصله اطمینان (CI): 0.78 تا 1.03؛ 11,250 شرکت‌کننده؛ 21 کارآزمایی/22 مقایسه؛ شواهد با کیفیت پائین) گزارش شد. خطر مطلق یک رویداد قلبی‌عروقی مهم 6.36% در گروه استنت‌های دارای پوشش دارویی در مقایسه با 6.63% در گروه استنت‌های فلزی بدون پوشش (RR: 0.96؛ 95% CI؛ 0.83 تا 1.11؛ 10,939 شرکت‌کننده؛ 19 کارآزمایی/20 مقایسه؛ شواهد با کیفیت بسیار پائین) بود. نتایج تجزیه‌وتحلیل مرحله‌‏ای کارآزمایی نشان داد که به اطلاعات کافی برای تایید یا رد کاهش 10% خطر نسبی پیش‌بینی شده در مورتالیتی به هر علتی یا وقایع قلبی‌عروقی جدی در حداکثر دوره پیگیری دسترسی نداریم.

متاآنالیزها در حداکثر دوره پیگیری، شواهد مثبتی را از مقایسه استنت‌های دارای پوشش دارویی و استنت‌های فلزی بدون پوشش در خطر بروز عارضه جانبی جدی نشان دادند. خطر مطلق یک عارضه جانبی جدی 18.04% در گروه استنت‌های دارای پوشش دارویی در مقایسه با 23.01% در گروه استنت‌های فلزی بدون پوشش (RR: 0.80؛ 95% CI؛ 0.74 تا 0.86؛ 11,724 شرکت‌کننده؛ 22 کارآزمایی/23 مقایسه؛ شواهد با کیفیت پائین) بود، و تجزیه‌وتحلیل مرحله‌‏ای کارآزمایی این نتایج را تاکید کرد. زمانی که هر نوع خاصی از عوارض جانبی وارد شده در پیامد عارضه جانبی جدی به‌طور جداگانه ارزیابی شدند، اکثر وقایع، بازسازی دوباره جریان خون (revascularisation) عروق هدف بودند. زمانی که بازسازی دوباره جریان خون عروق هدف به‌طور جداگانه‌ای تحلیل شدند، متاآنالیز شواهدی را از منفعت استنت‌های دارای پوشش دارویی نشان داد و تجزیه‌وتحلیل مرحله‌‏ای کارآزمایی این نتیجه را تایید کرد.

متاآنالیزها در حداکثر دوره پیگیری، هیچ تفاوتی را در مقایسه استنت‌های دارای پوشش دارویی و استنت‌های فلزی بدون پوشش بر خطر مورتالیتی قلبی‌عروقی (RR: 0.91؛ 95% CI؛ 0.76 تا 1.09؛ 9248 شرکت‌کننده؛ 14 کارآزمایی/15 مقایسه؛ شواهد با کیفیت بسیار پائین) یا انفارکتوس میوکارد (RR: 0.98؛ 95% CI؛ 0.82 تا 1.18؛ 10,217 شرکت‌کننده؛ 18 کارآزمایی/19 مقایسه؛ شواهد با کیفیت بسیار پائین) نشان ندادند. نتایج تجزیه‌وتحلیل مرحله‌‏ای کارآزمایی نشان داد که اطلاعات کافی را برای تایید یا رد کاهش 10% نسبت خطر پیش‌بینی شده در مورتالیتی قلبی‌عروقی و انفارکتوس میوکارد نداشتیم.

هیچ یک از کارآزمایی‌ها نتایجی را در خصوص کیفیت زندگی یا آنژین گزارش ندادند.

نتیجه‌گیری‌های نویسندگان

شواهد کنونی نشان می‌دهد که استنت‌های دارای پوشش دارویی ممکن است عوارض جانبی جدی کمتری در مقایسه با استنت‌های فلزی بدون پوشش داشته باشند، بدون اینکه خطر مورتالیتی به هر علتی یا وقایع مهم قلبی‌عروقی را افزایش دهند. با این حال، تجزیه‌وتحلیل مرحله‌‏ای کارآزمایی نشان داد که در حال حاضر اطلاعات کافی برای ارزیابی کاهش 10% خطر نسبی برای مورتالیتی به هر علتی، وقایع مهم قلبی‌عروقی، مورتالیتی قلبی‌عروقی، یا انفارکتوس میوکارد وجود ندارد، و هیچ اطلاعاتی در خصوص کیفیت زندگی یا آنژین در دست نبودند. کیفیت شواهد در این مطالعه مروری از پائین تا بسیار پائین متغیر بود، و ممکن است نتیجه واقعی از نتایج ارائه شده در این مطالعه مروری به‌طور قابل‌ توجهی متفاوت باشد.

انجام کارآزمایی‌های بالینی تصادفی‌سازی شده بیشتری با خطر پائین سوگیری و خطرات پائین خطاهای تصادفی مورد نیاز است تا مزایا و معایب استنت‌های دارای پوشش دارویی برای سندرم کرونری حاد به‌طور مناسبی ارزیابی شوند. داده‌های بیشتری در خصوص پیامدهای مورتالیتی به هر علتی، وقایع قلبی و عروقی مهم، کیفیت زندگی، و آنژین برای کاهش خطر خطای تصادفی مورد نیاز است.

PICO

Population

Intervention

Comparison

Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

خلاصه به زبان ساده

disponible en

مزایا و مضرات استنت‌های دارای پوشش دارویی در مقابل استنت‌‌های فلزی بدون پوشش در افراد مبتلا به سندرم کرونری حاد

پیشینه

سندرم کرونری حاد اصطلاحی است که برای حملات قلبی و نیز درد قفسه سینه در طول استراحت استفاده می‌شود. این سندرم به دلیل کاهش تامین خون مورد نیاز قلب بروز می‌کند که ناشی از باریک‌ شدن رگ‌های تامین کننده قلب یا یک لخته خونی است. طبق برآورد سازمان جهانی بهداشت، در سال 2012 حدود 3.7 میلیون نفر در اثر ابتلا به سندرم کرونری حاد فوت شدند. مداخله عروق کرونر از راه پوست، که عموما به آنژیوپلاستی کرونری شناخته شده، برای درمان عروق بسته یا باریک‌ شده قلب با باد کردن یک بالون استفاده می‌شوند تا به خون اجازه ‌دهند که دوباره در رگ جریان داشته باشد. سپس برای اطمینان از باز ماندن رگ، استنت‌های کرونری در آن قرار داده می‌شوند. دو نوع استنت برای استفاده در دسترس است، استنتی که با دارو پوشیده شده و به کاهش خطر بسته شدن رگ در آینده کمک می‌کند (استنت با پوشش دارویی) و استنت بدون پوشش (استنت فلزی بدون پوشش). هدف ما در این مطالعه مروری، ارزیابی مزایا و مضرات استنت‌های دارای پوشش دارویی در مقایسه با استنت‌های فلزی بدون پوشش بود.

ویژگی‌های مطالعه

بانک‌های اطلاعاتی را از ابتدا تا ژانویه 2017 جست‌وجو کرده و 25 کارآزمایی را یافتیم که در آنها افراد به‌طور تصادفی در گروه‌هایی قرار گرفتند که استنت‌های دارای پوشش دارویی و استنت‌های فلزی بدون پوشش را دریافت کردند. 25 کارآزمایی (26 مقایسه) شامل 12,503 بزرگسال با میانگین سنی 60.8 سال بودند.

نتایج کلیدی

افرادی که استنت‌های دارای پوشش دارویی دریافت کردند در مقایسه با بیمارانی که در گروه استنت‌های فلزی بدون پوشش قرار گرفتند، به نظر می‌رسید که با خطر کمتری از ابتلا به عارضه جانبی جدی روبه‌رو بودند، بدون اینکه بر خطر مرگ به هر دلیل یا به‌طور خاص خطر مرگ ناشی از بیماری قلبی، یا حمله قلبی تاثیر بگذارد. عارضه جانبی جدی که احتمالا با استنت‌های دارای پوشش دارویی پیشگیری می‌شود، تکرار آنژیوپلاستی بالون است. هیچ داده‌ای از کیفیت زندگی یا درد قفسه سینه (آنژین) پس از دریافت استنت گزارش نشد.

کیفیت شواهد

شواهد باید با احتیاط تفسیر شوند، زیرا به دلیل محدودیت‌های کارآزمایی‌های بالینی تصادفی‌سازی شده وارد شده، کیفیت را از پائین تا بسیار پائین قضاوت کردیم. کیفیت بسیار پائین اساسا به دلیل این بود که همه کارآزمایی‌های وارد شده در معرض خطر بالای سوگیری (bias) قرار داشتند. در نتیجه، نتایج ممکن است منفعت بیش‌تری را از استنت‌های دارویی نسبت به اثر واقعی آنها در بالین نشان دهند. به‌ علاوه، اغلب پیامدها در مطالعه مروری ما قدرت آماری کافی را نداشتند. در نتیجه انجام کارآزمایی‌های بالینی تصادفی‌سازی شده‌ای که به خوبی طراحی شده‌ باشند، ممکن است نتایج مذکور را تغییر دهند.

Authors' conclusions

Implications for practice

The current evidence suggests that drug‐eluting stents may lead to fewer serious adverse events compared with bare‐metal stents. When assessing each specific type of event included in the serious adverse event outcome separately, we found that the majority of events were target vessel revascularisation. The reduction in serious adverse events came without increasing the risk of all‐cause mortality or major cardiovascular events. Our Trial Sequential Analysis showed that there was currently not enough information to assess a risk ratio reduction of 10% for all‐cause mortality, major cardiovascular events, cardiovascular mortality, or myocardial infarction, and there were no data on quality of life or angina. The evidence for the above outcomes was of low to very low quality, and the true result may depart substantially from the results presented in this systematic review.

Implications for research

Our Trial Sequential Analysis showed that there was currently not enough information to assess a risk ratio reduction of 10% for all‐cause mortality, major cardiovascular events, cardiovascular mortality, and myocardial infarction. We therefore recommend that future randomised clinical trials assessing these outcomes be conducted, as currently the superiority of drug‐eluting stents relies on a possible reduced risk of target vessel revascularisation. In addition, we found no trials reporting quality of life or angina, therefore future trials should also assess these outcomes. Furthermore, trials conducted up to now have not comprehensively reported serious adverse events, and future trials should avoid continuing this practice. Future trials should be conducted with low risk of systematic error and low risk of random errors. Such trials ought to be designed and reported according to the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) and CONSORT guidelines (Chan 2013; Schulz 2010).

There is a need for systematic reviews assessing serious adverse events in quasi‐randomised and observational studies.

Summary of findings

Open in table viewer

Summary of findings 1. Drug‐eluting stents compared to bare‐metal stents for acute coronary syndrome

Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
Drug‐eluting stents compared to bare‐metal stents for acute coronary syndrome
Patient or population: People with acute coronary syndrome Setting: Hospital Intervention: Drug‐eluting stents Comparison: Bare‐metal stents
Outcomes	Risk with bare‐metal stents	Risk with drug‐eluting stents	Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
All‐cause mortality at maximum follow‐up Follow‐up: median 12 months	Study population		RR 0.90 (0.78 to 1.03)	11,250 (21 RCTs/22 comparisons)	⊕⊕⊝⊝ LOW ^1,2	Trial Sequential Analysis for a RRR of 10% showed that neither the boundary for futility, benefit or harm was breached, hence the risk of imprecision of the outcome result is high. Multiple eligible treatments were used in 1 trial, generating a further comparison (21 trials reporting on 22 experimental groups).
	78 per 1000	70 per 1000 (60 to 80)	RR 0.90 (0.78 to 1.03)	11,250 (21 RCTs/22 comparisons)	⊕⊕⊝⊝ LOW ^1,2
Serious adverse events at maximum follow‐up Follow‐up: median 12 months	Study population		RR 0.80 (0.74 to 0.86)	11,724 (22 RCTs/23 comparisons)	⊕⊕⊝⊝ LOW ³	Trial Sequential Analysis for a RRR of 10% showed that the boundary for benefit was breached, hence the risk of imprecision of the outcome result is low. Multiple eligible treatments were used in 1 trial, generating a further comparison (22 trials reporting on 23 experimental groups).
	230 per 1000	184 per 1000 (170 to 197)	RR 0.80 (0.74 to 0.86)	11,724 (22 RCTs/23 comparisons)	⊕⊕⊝⊝ LOW ³
Major cardiovascular events at maximum follow‐up Follow‐up: median 12 months	Study population		RR 0.96 (0.83 to 1.11)	10,939 (19 RCTs/20 comparisons)	⊕⊝⊝⊝ VERY LOW ^1,3	Trial Sequential Analysis for a RRR of 10% showed that neither the boundary for futility, benefit or harm was breached, hence the risk of imprecision of the outcome result is high. Multiple eligible treatments were used in 1 trial, generating a further comparison (19 trials reporting on 20 experimental groups).
	66 per 1000	63 per 1000 (55 to 73)	RR 0.96 (0.83 to 1.11)	10,939 (19 RCTs/20 comparisons)	⊕⊝⊝⊝ VERY LOW ^1,3
Quality of life at maximum follow‐up ‐ not reported	‐	‐	‐	‐	‐
Cardiovascular mortality at maximum follow‐up Follow‐up: median 12 months	Study population		RR 0.91 (0.76 to 1.09)	9248 (14 RCTs/15 comparisons)	⊕⊝⊝⊝ VERY LOW ^1,3	Trial Sequential Analysis for a RRR of 10% showed that neither the boundary for futility, benefit or harm was breached, hence the risk of imprecision of the outcome result is high. Multiple eligible treatments were used in 1 trial, generating a further comparison (14 trials reporting on 15 experimental groups).
	58 per 1000	53 per 1000 (44 to 63)	RR 0.91 (0.76 to 1.09)	9248 (14 RCTs/15 comparisons)	⊕⊝⊝⊝ VERY LOW ^1,3
Myocardial infarction at maximum follow‐up Follow‐up: median 12 months	Study population		RR 0.98 (0.82 to 1.18)	10,217 (18 RCTs/19 comparisons)	⊕⊝⊝⊝ VERY LOW ^1,3	Trial Sequential Analysis for a RRR of 10% showed that neither the boundary for futility, benefit or harm was breached, hence the risk of imprecision of the outcome result is high. Multiple eligible treatments were used in 1 trial, generating a further comparison (18 trials reporting on 19 experimental groups).
	48 per 1000	47 per 1000 (39 to 56)	RR 0.98 (0.82 to 1.18)	10,217 (18 RCTs/19 comparisons)	⊕⊝⊝⊝ VERY LOW ^1,3
Angina at maximum follow‐up ‐ not reported	‐	‐	‐	‐	‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; OR: odds ratio; RCT: randomised controlled trial; RR: risk ratio; RRR: risk ratio reduction
GRADE Working Group grades of evidence High quality: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
¹Downgraded one level for serious risk of imprecision due to our required information size not being met. ²Downgraded one level for serious risk of bias due to all trials being at high risk of bias, but as mortality is a more objective outcome, lack of blinding of participants, personnel and outcome assessors may not bias the outcome as much. ³Downgraded two levels for very serious risk of bias due to all trials being at high risk of bias.

Background

Description of the condition

Cardiovascular disease, of which ischaemic heart disease is by far the largest component, is considered to be the number one cause of death globally (Lloyd‐Jones 2010; Nichols 2014; Rosamond 2008). Ischaemic heart disease is generally divided into acute coronary syndrome and stable ischaemic heart disease (Roffi 2016). The disease remains prevalent, with more than 7 million people worldwide expected to develop acute coronary syndrome each year (White 2008). According to the World Health Organization, 7.4 million people died from ischaemic heart disease in 2012, representing 15% of all global deaths, with acute coronary syndrome accounting for approximately half of the deaths (Turpie 2006; WHO 2015). The in‐hospital mortality proportions across different countries range from 5% to 10% (Gupta 2003). Ischaemic heart disease remains increasingly prevalent and costly to treat due to an increase in life expectancy and a decrease in death rates (Cooper 2000; Schmidt 2012).

Acute coronary syndrome is a collective term for the following:

unstable angina pectoris (myocardial ischaemia at rest or minimal exertion in the absence of cardiomyocyte necrosis) (Roffi 2016);
non‐ST‐segment elevation myocardial infarction (NSTEMI); and
ST‐segment elevation myocardial infarction (STEMI) (Amsterdam 2014). Myocardial infarction can be recognised by clinical features, electrocardiographic (ECG) findings, elevated values of biochemical markers (biomarkers) of myocardial necrosis, imaging, or pathology (Thygesen 2012). The diagnosis of myocardial infarction is dependent on an elevation of the serum levels of cardiac‐specific troponin I, troponin T, or the myocardial band isoenzyme of creatine kinase (CK‐MB), among others (Roffi 2016). Cardiac troponin levels will usually be positive within one hour in people with myocardial infarction when using high‐sensitivity assays (Roffi 2016). If the initial test is negative, a repeat test is advisable after three hours to rule out a myocardial infarction, since in some cases there is a delay in the elevation of cardiac enzymes (Roffi 2016). The changes in the ST‐segment reflected in an ECG provide the basis for the distinction between NSTEMI and STEMI (Roffi 2016). Compared to NSTEMI and STEMI, unstable angina pectoris presents without the characteristic rise in cardiac‐specific biomarkers (Roffi 2016).

For research purposes, the causes of myocardial infarction are generally divided into five main classes (Thygesen 2012).

Type 1: spontaneous myocardial infarction related to atherosclerotic plaque rupture, ulceration, fissuring, erosion, or dissection with resulting intraluminal thrombus in one or more of the coronary arteries often caused by coronary artery disease.
Type 2: myocardial infarction secondary to an ischaemic imbalance, such as coronary artery spasm, coronary embolism, anaemia, arrhythmias, hypertension, or hypotension.
Type 3: myocardial infarction with symptoms suggestive of myocardial ischaemia and resulting in sudden unexpected cardiac death when biomarker values are unavailable or unobtainable before death.
Type 4a: myocardial infarction associated with percutaneous coronary intervention.
Type 4b: myocardial infarction associated with stent thrombosis, as documented by angiography or at autopsy.
Type 5: myocardial infarction associated with coronary artery bypass graft.

Major complications associated with myocardial infarction

Life‐threatening ventricular arrhythmias caused by changes in the electrophysiologic characteristics of the myocyte, electrolyte imbalance, continuous ischaemia, and variations in heart rate, which are all due to obstruction and hence reduced flow to the myocardium and myocardial necrosis (Brieger 2009; Stevenson 1989).
Mechanical complications caused by necrosis of the myocardium, such as ventricular wall rupture, septum rupture, and papillary muscle rupture (Brieger 2009; Pohjola‐Sintonen 1989; Stevenson 1989).
Cardiogenic shock caused by failure of the ventricle to pump an adequate amount of blood leading to systemic hypotension (Brieger 2009; Stevenson 1989).
Acute decompensated heart failure caused by impairment in systolic and diastolic function due to myocardial ischaemia (Brieger 2009).
Depression (Thombs 2006).

A narrowing of a coronary vessel causes unstable angina due to one of five reasons:

non‐occlusive thrombus on pre‐existing plaques;
dynamic obstruction, i.e. coronary vasoconstriction;
progressive mechanical obstruction, such as restenosis after percutaneous coronary intervention;
inflammation or infection; and
secondary unstable angina due to conditions increasing the oxygen demand, such as hypertension, thyrotoxicosis, and tachycardia (Braunwald 1998; Roffi 2016). Unstable angina is associated with lower mortality compared to myocardial infarction but similar rates of rehospitalisations; however, the patient may have a worse quality of life (Dudas 2013; Maddox 2007; Roffi 2016). People with unstable angina are also at risk of cardiogenic shock (Ruiz‐Bailén 2008).

Historically, the degree of luminal stenosis and the number of coronary arteries involved (single‐, double‐, or triple‐vessel disease) have defined the severity of ischaemic heart disease (Ringqvist 1983). Researchers have since developed more comprehensive scoring systems (Gensini 1983; Seizer 1982). Coronary angiography score and two additional scores, that is vascular scoring and stenosis scoring, determine the Gensini score (Gensini 1983). The results of the coronary angiography determine the SYNTAX score, which takes into account lesion complexity, lesion location, and number of lesions (Sianos 2005). The SYNTAX score II is used to improve the decision‐making in choosing between percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) for a long‐term, individualised risk assessment in people with complex ischaemic heart disease. The SYNTAX score II combines the anatomical‐based SYNTAX score (Sianos 2005), as well as seven clinical variables (creatinine clearance, peripheral vascular disease, unprotected left main coronary disease, sex, chronic obstructive pulmonary disease, age, and left ventricular ejection fraction) (Farooq 2013; Sianos 2005).

A glossary of medical terms is in Appendix 1.

Description of the intervention

Percutaneous coronary intervention is performed by inserting an access sheath into a peripheral artery (most often the femoral artery or the radial artery); a catheter is advanced, under X‐ray screening, through the blood vessels to the aortic root, at the origin of the coronary arteries. Other tools such as balloons and stents can then be advanced down the artery, over a guide wire, to the location of the narrowing or blockage (Cantor 2005; Hamon 2009). Andreas Roland Grüntzig performed the first percutaneous coronary intervention in 1977 (Grüntzig 1978). Percutaneous coronary intervention has since then evolved to become one of the cornerstones in the treatment of ischaemic heart disease.

Percutaneous coronary intervention in people with acute coronary syndrome can be performed both as primary percutaneous coronary intervention (primary PCI) and secondary percutaneous coronary intervention (secondary PCI) (Meyer 1981; Meyer 1982; Meyer 1982a; Meyer 1983). Primary PCI may be performed in the acute phase of myocardial infarction if patients present with STEMI. Patients presenting with NSTEMI or unstable angina pectoris that is characterised as high‐risk acute coronary syndrome may also benefit from undergoing angiography and percutaneous coronary intervention either immediately or as part of an early (within 24 hours) invasive strategy (Hoenig 2010; Roffi 2016). The high‐risk factors include haemodynamic instability, severe left ventricular dysfunction, recurrent or persistent rest angina, mechanical complications, sustained ventricular tachycardia, and dynamic ST‐T wave electrocardiographic changes (Hoenig 2010).

Secondary PCI includes both subacute PCI and elective PCI. Subacute PCI is generally performed within the first 72 hours after symptom debut in people with unstable angina pectoris or NSTEMI who are not candidates for primary PCI because of being haemodynamically stable with medical therapy and not at high risk (Breall 2016; Roffi 2016). Elective PCI is performed in patients where coronary artery bypass grafting is not indicated, as well as in patients who are dissatisfied with their quality of life because of symptoms related to ischaemic heart disease or with adverse events due to their medical treatment (Levin 2016).

The first percutaneous coronary intervention was performed by inflating a balloon at the blockage of the coronary artery to dilate the artery ('balloon angioplasty') (Grüntzig 1978; Grüntzig 1979). The healing properties of the treatment seem to be related to the percutaneous coronary intervention expanding the outer diameter of the blocked coronary artery and not by decompression of the arterial plaque (Düber 1986). Balloon angioplasty, however, generally did not seem to achieve a long‐lasting result, with restenosis occurring over time (Dangas 2002; Puel 1988). In an attempt to keep the lumen open longer, the next development was a small metallic scaffold called a 'stent'. These tubular devices are expanded over a balloon and press against the walls of the artery to keep it open (Puel 1988). These first devices were subsequently termed 'bare‐metal stents' (after the later introduction of 'drug‐eluting stents') and had improved outcomes over balloon angioplasty, but were still associated with high rates of restenosis over time (Cutlip 2002; Erbel 1998; Fischman 1994; Macaya 1996; Puel 1988; Serruys 1994; Serruys 1998). A Cochrane review from 2005 found that primary stenting versus balloon angioplasty alone seemed to reduce the rate of reinfarction and target vessel revascularisation (Nordmann 2005).

The next stents were drug‐eluting stents consisting of three main components:

a metal mesh;
an antiproliferative drug (e.g. sirolimus, paclitaxel, zotarolimus, or everolimus); and
a polymer used to coat the metal mesh.

Antiproliferative drugs to limit the excessive growth of neointima using cytotoxic or cytostatic agents, as well as a polymer to control the release of the antiproliferative drug, supplemented the metal mesh (Degertekin 2002; Fajadet 2006; Holmes 2004; Lee 2005; Morice 2002; Moses 2003; Stone 2004; Stone 2004a).

Newer drug‐eluting stents have decreased strut thickness and are meant to have improved flexibility/deliverability, enhanced polymer biocompatibility/drug‐eluting profiles, and superior re‐endothelialisation kinetics (Serruys 2010; Stone 2010). They typically use everolimus or zotarolimus as their antiproliferative drug (Serruys 2010; Stone 2010).

The polymer‐coating of the drug‐eluting stents has, however, been linked with adverse events, such as stent thrombosis (Chen 2015). Both drug‐eluting stents with a biodegradable polymer as well as polymer‐free drug‐eluting stents have therefore been developed. Polymer‐free drug‐eluting stents use the same antiproliferative drugs (such as paclitaxel or sirolimus) as the polymer drug‐eluting stents (Abizaid 2010; Chen 2015).

In an attempt to further reduce the risk of restenosis, bioresorbable (also called biodegradable) stents were developed. The principal components of the bioresorbable stent are the same as the drug‐eluting stents; in most cases, a polylactic acid mesh replaces the metal mesh (Haude 2013; Puricel 2015). The polylactic acid mesh is broken down and removed over time (Tamai 2000). The same types of drugs (everolimus, paclitaxel, sirolimus) used in drug‐eluting stents along with Biolimus are used in bioresorbable stents (Haude 2013; Haude 2016; Puricel 2015).

Bare‐metal stents, drug‐eluting stents, and bioresorbable stents are used in modern percutaneous coronary interventions, with drug‐eluting stents generally being the first choice (Windecker 2014). Guidelines recommend that acute coronary syndrome patients receive 12 months of antiplatelet therapy (aspirin plus a G protein‐coupled purinergic (P2Y12 receptor) blocker) regardless of whether percutaneous coronary intervention is performed (Windecker 2014). The recommended minimum length of duration of antiplatelet therapy for the implant of the bare‐metal stents and drug‐eluting stents is one month and six months, respectively (Windecker 2014).

Adverse events associated with percutaneous coronary intervention include death, coronary artery complications (such as perforation of the artery, distal embolisation (passage of an intravascular mass, which is capable of clogging capillaries), or stent thrombosis), myocardial infarction (type 4 myocardial infarction) (Thygesen 2012), vascular complications (such as bleeding or infection at the access site, retroperitoneal bleeding, or atheroembolism), stroke, and acute kidney failure (Baim 1996; Cantor 1998; Stankovic 2004).

How the intervention might work

Percutaneous coronary intervention aims to decrease the stenosis of the coronary artery resulting in increased blood flow to the myocardium of the heart, which is thought to limit ischaemia and potentially reinfarction. Drug‐eluting stents may be more beneficial than bare‐metal stents because they release antiproliferative drugs, which may limit neointimal growth (Degertekin 2002; Fajadet 2006; Holmes 2004; Lee 2005; Morice 2002; Moses 2003; Stone 2004; Stone 2004a). The new bioresorbable stents as well as the polymer‐free drug‐eluting stents may be even more beneficial since they remove material that has been associated with adverse events (Abizaid 2010; Chen 2015; Haude 2013; Puricel 2015).

Why it is important to do this review

The prevalence of ischaemic heart disease is considerable and causes one‐third of all deaths in people over 35 years of age (Lloyd‐Jones 2010; Nichols 2014; Rosamond 2008). Beneficial treatments can therefore alleviate a considerable disease burden and healthcare cost. Former reviews have found conflicting evidence of the efficacy and safety of drug‐eluting stents versus bare‐metal stents.

A 2010 Cochrane review compared drug‐eluting stents versus bare‐metal stents in participants with both acute coronary syndrome and stable ischaemic heart disease (Greenhalgh 2010). The authors found no significant difference on all‐cause mortality, incidence of acute myocardial infarction, or thrombosis. However, the review showed indications of beneficial effects of drug‐eluting stents compared with bare‐metal stents on target lesion revascularisation, target vessel revascularisation, and a composite outcome of cardiac events.

Six non‐Cochrane reviews compared drug‐eluting stents with bare‐metal stents in participants with ischaemic heart disease (De Luca 2012; Kastrati 2007; Roukoz 2009; Stettler 2007; Suh 2011; Zheng 2014). Three reviews included only STEMI patients (De Luca 2012; Suh 2011; Zheng 2014). Two of these reviews found no difference between drug‐eluting stents versus bare‐metal stents on all‐cause mortality and stent thrombosis, but a beneficial effect on target vessel revascularisation (De Luca 2012; Suh 2011). Suh 2011 observed improvement of recurrent myocardial infarction with drug‐eluting stents, while De Luca 2012 found no effect. Zheng 2014, which included only four trials, compared the effects of bare‐metal stents with drug‐eluting stents in STEMI participants at five‐year follow‐up. The review found that there was no difference in the interventions with regard to all‐cause mortality and acute myocardial infarction, as well as no effect on thrombosis, except for its occurrence later than one year after percutaneous coronary intervention. Three reviews assessed the effects of drug‐eluting stents versus bare‐metal stents in participants with ischaemic heart disease (including STEMI, NSTEMI, unstable angina pectoris, and stable angina pectoris) (Kastrati 2007; Roukoz 2009; Stettler 2007). Of the three reviews, two compared sirolimus stents and paclitaxel stents with bare‐metal stents (Roukoz 2009; Stettler 2007), while one compared only sirolimus stents with bare‐metal stents (Kastrati 2007). All reviews found no statistically significant difference between drug‐eluting stents and bare‐metal stents on all‐cause mortality. While Stettler 2007 found a beneficial effect in favour of the drug‐eluting stents using sirolimus on myocardial infarction, they found no effect for drug‐eluting stents using paclitaxel. Kastrati 2007 and Roukoz 2009 found no difference between drug‐eluting stents versus bare‐metal stents on myocardial infarction. All three reviews found a beneficial effect favouring drug‐eluting stents on target vessel revascularisation (Kastrati 2007; Roukoz 2009; Stettler 2007).

Current guidelines on drug‐eluting stents versus bare‐metal stents for acute coronary syndrome

The 2011 American College of Cardiology Foundation/American Heart Association/Society for Cardiovascular Angiography and Interventions guideline recommends drug‐eluting stents as a useful alternative to bare‐metal stents to prevent restenosis in cases where there is an increased risk of restenosis, and the patient is likely to be able to tolerate and comply with prolonged dual antiplatelet therapy (so‐called evidence A for STEMI, evidence C for NSTEMI/unstable angina pectoris; evidence levels are explained in the glossary in Appendix 1) (Levine 2011). The clinical situations associated with increased risk of restenosis are left main disease, small vessels, in‐stent restenosis, bifurcations, diabetes, long lesions, multiple lesions, and saphenous vein grafts (Levine 2011; Levine 2016). The guideline also states that bare‐metal stents should be used in people with a high risk of bleeding; inability to comply with one year of dual antiplatelet therapy; or anticipated invasive or surgical procedures in the next year, which the 2013 American College of Cardiology Foundation/American Heart Association Guideline for the Management of ST‐Elevation Myocardial Infarction also recommends (ACEP 2013; Levine 2011).

The National Institute for Health and Care Excellence (NICE) has assessed the effects and cost‐effectiveness of drug‐eluting stents compared with bare‐metal stents (NICE 2008). NICE recommends drug‐eluting stents in people where the target artery for treatment has less than a 3 mm calibre or the lesion is longer than 15 mm, and the price difference between drug‐eluting stents and bare‐metal stents is no more than GBP 300.

This review is an update of the 2010 Cochrane review mentioned above (Greenhalgh 2010), which has now been divided into a review including acute coronary syndrome participants and a review including stable ischaemic heart disease participants. With the present review we planned to do the following:

take full account of the risk of systematic errors ('bias'), design errors, and risks of random errors ('play of chance') (Higgins 2011; Jakobsen 2014; Keus 2010; Thorlund 2011; Wetterslev 2008);
include trials irrespective of outcome, follow‐up duration, and number of participants;
assess outcomes at several time points and take into account the variability of the follow‐up period; and
include all types of drug‐eluting stents, including polymer‐free stents and bioresorbable stents.

Objectives

To assess the benefits and harms of drug‐eluting stents versus bare‐metal stents in people with acute coronary syndrome.

Methods

Criteria for considering studies for this review

Types of studies

We searched for randomised clinical trials (both individual and cluster‐randomised trials) irrespective of publication type, publication status, publication date, and language.

Types of participants

We included participants of any age with a diagnosis of acute coronary syndrome (according to the definition of the trialists).

Types of interventions

We included any type of drug‐eluting stents, including bioresorbable stents and polymer‐free drug‐eluting stents.

We accepted any type of medical therapy as a co‐intervention to percutaneous coronary intervention.

We included any type of bare‐metal stent as a control intervention.

Types of outcome measures

Primary outcomes

All‐cause mortality.
Serious adverse events, defined as any untoward medical occurrence that resulted in death, was life‐threatening, was persistent, or led to significant disability, prolonged hospitalisation, or any medical event that had jeopardised the participant or required intervention to prevent it (ICH‐GCP 1997).
Major cardiovascular event, defined as a composite outcome consisting of cardiovascular mortality and myocardial infarction.
Quality of life measured on any valid scale, such as the Seattle Angina Questionnaire or 36‐Item Short Form Health Survey (SF‐36) (Ware 1992; Wyrwich 2004).

Secondary outcomes

Cardiovascular mortality (defined by the trialists).
Myocardial infarction (defined by the trialists).
Angina on a continuous scale, such as 'angina stability' and 'angina frequency' used in the Seattle Angina Questionnaire (Wyrwich 2004).

Exploratory outcomes

Stent thrombosis.
Target vessel revascularisation (defined by the trialists). In general, target vessel revascularisation is any repeat percutaneous intervention or surgical bypass of any segment of the target vessel (Hicks 2010).

We conducted meta‐analyses when possible of all dichotomous and continuous outcomes at the following two time points:

outcomes assessed at maximal follow‐up (the time point of primary interest); and
outcomes assessed at one month.

Search methods for identification of studies

Electronic searches

We identified trials through systematic searches of the following bibliographic databases:

the Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library (2017, Issue 1 of 12), 18 January 2017;
Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily, and MEDLINE (Ovid, 1946 to 18 January 2017);
Embase (Ovid, 1980 to 2017 week 3), 18 January 2017;
LILACS (Latin American and Caribbean Health Sciences Literature) (Bireme, 1982 to 19 January 2017);
SCI‐EXPANDED (Thomson Reuters, 1900 to 19 January 2017);
BIOSIS Web of Science (Thomson Reuters, 1926 to 19 January 2017).

We adapted the search strategy for MEDLINE Ovid to the other databases (Appendix 2). We applied the Cochrane sensitivity‐maximising randomised controlled trial filter to MEDLINE Ovid, adaptations of which were applied to the other databases, except CENTRAL and LILACS (Lefebvre 2011). We searched all databases from their inception and imposed no restriction on language of publication.

We also conducted a search of ClinicalTrials.gov (www.clinicaltrials.gov, searched on 30 March 2017), the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (www.apps.who.int/trialsearch, searched on 31 March 2017), European Medicines Agency (www.ema.europa.eu/ema/, searched on 3 April 2017), and the US Food and Drug Administration (www.fda.gov, searched on 4 April 2017). We also searched Google Scholar manually for trials not found in the preliminary search (Lefebvre 2011). We adapted the preliminary search strategy for MEDLINE (Ovid) for use in these databases.

We also searched the websites of the following pharmaceutical companies for ongoing or unpublished trials: Johnson & Johnson (www.jnj.com/about-jnj, searched on 30 March 2017), Medtronic (www.medtronic.com/us-en/index.html, searched on 30 March 2017), Boston Scientific (www.bostonscientific.com/en-US/Home.html, searched on 30 March 2017), and Abbott (www.abbott.com/, searched 30 March 2017).

Searching other resources

We identified additional trials from the reference lists of review articles and included trials.

Data collection and analysis

We performed the review following the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We conducted the analyses using Review Manager 5 (RevMan 2014), Stata 14 (Stata 2015), and Trial Sequential Analysis (CTU 2011).

Selection of studies

Two review authors (JF and EEN) independently screened titles and abstracts of all of the potentially eligible trials for inclusion. We coded all of these studies as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. In case of disagreement, we asked a third review author to arbitrate (JCJ). We retrieved the full‐text trial reports/publications, and two review authors (JF and EEN) independently screened the full texts and identified trials for inclusion. We reported reasons for exclusion of the ineligible studies. Any disagreements were resolved through discussion. We identified and excluded duplicates and collated multiple reports of the same trial so that each trial, rather than each report, was the unit of interest in this review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' tables.

Data extraction and management

We used a data collection form that we had piloted on three trials in the review to collect trial characteristics and outcome data. Two review authors (JF and EEN) independently extracted all trial characteristics from the included trials. In case of disagreement, we asked a third review author to arbitrate (JCJ). We extracted the following trial characteristics.

Methods: duration of the trial, details of any 'run‐in' period, and date of publication.
Participants: number randomised, number analysed, number lost to follow up/withdrawn, mean age, sex, inclusion criteria, and exclusion criteria.
Interventions: intervention, comparison, concomitant medications, and excluded medications.
Outcomes: primary and secondary outcomes specified and collected and time points reported.
Notes: trial funding and notable conflicts of interest of the trial authors.

Two review authors (JF and EEN) independently extracted outcome data from the included studies. Any disagreements were resolved by consensus or by involving a third review author (JCJ). One review author (JF) transferred the data into the Review Manager 5 file (RevMan 2014). We double‐checked that data were entered correctly by comparing the data presented in the review with the study reports. A second review author (EEN) spot‐checked study characteristics for accuracy against the trial report.

Assessment of risk of bias in included studies

Two review authors (JF and EEN) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Any disagreements were resolved by discussion or by involving a third review author (JCJ). We assessed the risk of bias according to the random sequence generation, allocation sequence concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other potential sources of bias. These 'risk of bias' domains enable the classification of randomised trials with an overall bias assessment at low or high risk of bias. Trials with one domain at either high or unclear risk of bias are considered at overall high risk of bias as previous meta‐epidemiological studies demonstrate that these two categories do not differ in their overestimation of benefits and underestimation of harms (Gluud 2006; Kjaergard 2001; Lundh 2012; Lundh 2017; Moher 1998; Savović 2012; Savović 2012a; Schulz 1995; Wood 2008). For additional details on how we assessed the risk of bias, see Appendix 3. We assessed the risk of bias at both trial level and outcome level.

Assessment of bias in conducting the review

We conducted the review according to the published protocol and have reported any deviations from it in the Differences between protocol and review section of this review.

Measures of treatment effect

Dichotomous outcomes

We calculated risk ratios (RR) with 95% confidence intervals (CIs) for dichotomous outcomes.

Continuous outcomes

We had planned to use end scores in our analyses if both end scores and change scores were reported. If only change scores were reported, we would use these, but no trials reported any continuous outcomes. We planned to calculate mean differences (MD) and standardised mean differences (SMD) with 95% CIs for continuous outcomes, as well as Trial Sequential Analysis‐adjusted CIs. We planned to use the SMD when all trials assessed the same outcome but measured it in a variety of ways, for example with different scales (Higgins 2011).

Unit of analysis issues

Where multiple trial arms were reported in a single trial, we only included the relevant arms. In the case of a trial with two relevant experimental arms (PASEO 2009a; PASEO 2009b), we included the two arms as different comparisons and halved the control group to avoid double‐counting.

Dealing with missing data

We contacted investigators or study sponsors to obtain any missing data.

Dichotomous outcomes

We did not impute missing values for any outcomes in our primary analysis. We imputed data in two of our sensitivity analyses (see Sensitivity analysis).

Continuous outcomes

We did not impute missing values for any outcomes in our primary analysis. We planned to impute data in our sensitivity analysis for dichotomous and continuous outcomes (see Sensitivity analysis).

If studies did not report standard deviations (SD), we planned to calculate them using data from the trial where possible.

Assessment of heterogeneity

We primarily investigated forest plots to visually assess for any sign of heterogeneity. We additionally assessed the presence of statistical heterogeneity by Chi² test (threshold P < 0.10) and measured the quantities of heterogeneity using the I² statistic (Higgins 2002; Higgins 2003). We followed the recommendations for thresholds in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011):

0% to 40%: might not be important;
30% to 60%: may represent moderate heterogeneity;
50% to 90%: may represent substantial heterogeneity; and
75% to 100%: considerable heterogeneity.

We investigated possible heterogeneity through subgroup analyses. Ultimately, we performed meta‐analyses of our outcomes (Higgins 2011).

Assessment of reporting biases

We planned to assess publication bias and other reporting biases by visual inspection of funnel plots for our primary outcomes if we included at least 10 trials (Higgins 2011). Using the asymmetry of the funnel plot, we assessed the risk of bias. To attempt to correct for the impact of any reporting bias, we used the 'trim and fill' method (Duval 2000; Taylor 1998).

For dichotomous outcomes, we tested asymmetry with the Harbord test, Harbord 2006, if Tau² was less than 0.1 and with the Rücker test, Rücker 2008, if Tau² was more than 0.1.

Data synthesis

Meta‐analysis

We undertook this systematic review according to the recommendations provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), according to Keus 2010, and according to the eight‐step assessment suggested by Jakobsen 2014. We used the statistical software Review Manager 5, provided by Cochrane, to meta‐analyse data (RevMan 2014).

We used Stata, Stata 2015, in case of zero‐event trials where Review Manager 5's zero‐event handling (replacing zero with a constant of 0.5) was not sufficient, for example in cases with a skewed number of participants between groups, we planned to handle this with reciprocal zero‐event handling according to Sweeting 2004. Additionally, we planned on using Stata 2015 in case we needed to undertake meta‐regression (post hoc).

We used Trial Sequential Analysis to assess and control the risk of random error (see below). If a meta‐analysis did not reach the required information size, we presented Trial Sequential Analysis‐adjusted confidence intervals to account for the lack of information.

We included all studies in our initial analyses and planned to conduct a sensitivity analysis of trials at low risk of bias, however no such trials were found. If the results were similar, we planned to base our primary conclusions on the overall analysis. If results differed, we planned to base our primary conclusions on trials at low risk of bias.

Trial Sequential Analysis

Cumulative meta‐analyses are at risk of producing random errors due to sparse data and multiple testing of accumulating data (Brok 2008; Brok 2009; Higgins 2011a; Pogue 1997; Thorlund 2009; Wetterslev 2008); Trial Sequential Analysis can be applied to control this risk (CTU 2011; Jackson 2017; Thorlund 2011; Wetterslev 2017). The required information size (i.e. the number of participants needed in a meta‐analysis to detect or reject a certain intervention effect) can be calculated in order to minimise random errors (Wetterslev 2008; Wetterslev 2009). The required information size takes into account the event proportion in the control group, the assumption of a plausible risk ratio reduction, and the heterogeneity of the meta‐analysis (Turner 2013; Wetterslev 2008; Wetterslev 2009; Wetterslev 2017). Trial Sequential Analysis enables testing for significance each time a meta‐analysis includes a new trial. On the basis of the required information size, trial sequential monitoring boundaries can be constructed, enabling the determination of the statistical inference concerning cumulative meta‐analysis that has not yet reached the required information size (Wetterslev 2008; Wetterslev 2017).

Firm evidence for benefit or harms may be established if the trial sequential monitoring boundary is crossed before reaching the required information size, in which case further trials may turn out to be superfluous. In contrast, if the boundaries for benefit or harm are not surpassed, it may be established that it is necessary to continue with further trials before a certain intervention effect can be detected or rejected. Firm evidence for lack of the postulated intervention effect, which occurs when the cumulative Z‐score crosses the trial sequential monitoring boundaries for futility, can also be assessed with Trial Sequential Analysis.

To control the risks of random error, we used relatively conservative estimations of the anticipated intervention effect estimates (Jakobsen 2014). Large anticipated intervention effects lead to small required information sizes, and the thresholds for significance will be too lenient (Jakobsen 2014).

We analysed all primary and secondary outcomes with Trial Sequential Analysis. We used the following assumptions.

Primary outcomes

We estimated the diversity‐adjusted required information size based on the proportion of participants with an outcome in the control group (Wetterslev 2009). We used a risk of a type I error of 2% (Jakobsen 2014), a risk of a type II error of 20%, and the diversity suggested by the trials in the meta‐analysis (Jakobsen 2014). We used a risk of a type I error of 2% due to us including four primary outcomes, hence adjusting our type I error accordingly (Jakobsen 2014). We chose a risk of a type II error of 20% based primarily on tradition; it could be argued this value should be lower (Castellini 2016).

As anticipated intervention effects we used the following risk ratio reductions (RRR) because they seemed to be the maximum realistic intervention effect estimates based on former studies, trials, and meta‐analyses.

All‐cause mortality: RRR of 10% (De Luca 2012; Suh 2011).
Serious adverse events: RRR of 10%.
Major cardiovascular event defined as a composite outcome consisting of cardiovascular mortality and myocardial infarction: RRR of 10% (De Luca 2012; Suh 2011).
Quality of life measured on any valid scale, such as the Seattle Angina Questionnaire or SF‐36: we planned to use the observed SD, a clinically relevant mean difference equal to SD/2.

Secondary outcomes

We estimated the diversity‐adjusted required information size based on the proportion of participants with an outcome in the control group (Wetterslev 2009). We used a risk of a type I error of 2.5% (Jakobsen 2014), a risk of a type II error of 20%, and the diversity suggested by the trials in the meta‐analysis (Jakobsen 2014). We used a risk of a type I error of 2.5% due to us including three secondary outcomes, hence adjusting our type I error accordingly (Jakobsen 2016). We chose a risk of a type II error of 20% based primarily on tradition; it could be argued this value should be lower (Castellini 2016).

As anticipated intervention effects for the secondary outcomes in the Trial Sequential Analysis, we used the following RRRs because we assumed that these intervention effects were realistic based on former studies, trials, and meta‐analyses as cited below.

Myocardial infarction: RRR of 10% (De Luca 2012; Suh 2011).
Cardiovascular mortality: RRR of 10% (De Luca 2012; Suh 2011).
Angina (continuous outcome): the observed SD, a clinically relevant mean difference equal to SD/2.

Exploratory outcomes

We estimated the diversity‐adjusted required information size based on the proportion of participants with an outcome in the control group (Wetterslev 2009). We used a risk of a type I error of 3.33% (Jakobsen 2014), a risk of a type II error of 20%, and the diversity suggested by the trials in the meta‐analysis (Jakobsen 2014). We used a risk of a type I error of 3.33% due to us including two exploratory outcomes, hence adjusting our type I error accordingly (Jakobsen 2016). We chose a risk of a type II error of 20% based primarily on tradition; it could be argued this value should be lower (Castellini 2016).

As anticipated intervention effects for the exploratory outcomes in the Trial Sequential Analysis, we used the following RRRs because we assumed that these intervention effects were realistic based on former studies, trials, and meta‐analyses as cited below.

Stent thrombosis: RRR of 10% (De Luca 2012; Suh 2011).
Target vessel revascularisation: RRR of 30% (De Luca 2012; Suh 2011).

As supplementary Trial Sequential Analysis, we planned to use the limit of the confidence interval closest to zero effect as the anticipated intervention effect for all trial sequential analyses (Jakobsen 2014).

Assessment of significance

We assessed our intervention effects with both random‐effects meta‐analyses, DerSimonian 1986, and fixed‐effect meta‐analyses, Demets 1987, using the more conservative point estimate, that is the estimate closest to zero effect, of the two (Jakobsen 2014). If the two estimates were equal, we used the estimate with the widest confidence interval. We had four primary outcomes and three secondary outcomes, and therefore considered a P value less than 2% as significant for the primary outcomes and a P value less than 2.5% as significant for the secondary outcomes (Jakobsen 2014; Jakobsen 2016). We used the eight‐step procedure to assess if the thresholds for significance were crossed or not (Jakobsen 2014).

'Summary of findings' tables

We used the GRADE approach to assess the quality of the body of evidence associated with each of the primary outcomes (all‐cause mortality, serious adverse events, major cardiovascular events, and quality of life) and secondary outcomes (cardiovascular mortality, myocardial infarction, and angina) in our review (Guyatt 2008), constructing 'Summary of findings' tables using GRADEpro software (www.gradepro.org). The GRADE approach appraises the quality of a body of evidence based on the extent to which one can be confident that an estimate of effect or association reflects the item being assessed. The quality measure of a body of evidence considers within‐study risk of bias, directness of the evidence, heterogeneity of the data, precision of effect estimates, and risk of publication bias (Guyatt 2008). We assessed the precision of the effect estimates according to Jakobsen and colleagues (Jakobsen 2014). We included all studies in our primary analyses, as we found no trials at low risk of bias or at low risk of bias in all domains other than 'blinding of participants and personnel' (Gluud 2006; Kjaergard 2001; Lundh 2012; Moher 1998; Savović 2012; Schulz 1995; Wood 2008).

Subgroup analysis and investigation of heterogeneity

We performed the following subgroup analyses.

Type of drug‐eluting stents used:
- paclitaxel‐eluting stents;
- sirolimus‐eluting stents;
- zotarolimus‐eluting stents;
- everolimus‐eluting stents;
- bioresorbable stents;
- polymer‐free drug‐eluting stents; and
- mixed drug‐eluting stents.
Unstable angina pectoris/non‐ST‐segment elevation myocardial infarction participants compared with ST‐segment elevation myocardial infarction participants.
Length of maximum follow‐up:
- less than or equal to six months;
- between six months and 12 months;
- between one year and three years; and
- more than or equal to three years.
Comparison of the effect of drug‐eluting stents versus bare‐metal stents between trials with different clinical trial registration status:
- pre‐registration;
- postregistration; and
- no registration.

We only used the primary outcomes in our subgroup analyses.

We used the formal test for subgroup differences in Review Manager 5 (RevMan 2014).

We did not perform the following subgroup analyses due to lack of data.

Participants with diabetes compared with participants without diabetes.
Participants with high risk of bleeding (as defined by trialists) compared with participants without high risk of bleeding.
Age of participants:
- age 0 to 18 years;
- age 19 to 75 years; and
- age 76 years or above.

Sensitivity analysis

We planned to assess the potential impact of bias by performing a sensitivity analysis where we excluded trials at high risk of bias. However, we found no trials at low risk of bias.

As a secondary sensitivity analysis, we planned to only include trials with low risk of bias in all domains except 'blinding of participants and personnel', as we did not expect to find any trials at low risk of bias in this domain. However, we found no such trial. We thoroughly discussed the limitations of trials at high risk of bias in the Quality of the evidence.

To assess the potential impact of the missing data for dichotomous outcomes, we performed the two following analyses.

'Best‐worst‐case' scenario: we assumed that all participants lost to follow‐up in the experimental group survived, had no serious adverse event, had no major cardiovascular event, had no stent thrombosis, and had no target vessel revascularisation. We assumed that they also had a beneficial event with regard to quality of life and angina, defined as the group mean plus both one and two SDs of the group mean (Jakobsen 2014). We assumed that all of those with missing outcomes in the control group died, had a serious adverse event, had a major cardiovascular event, had stent thrombosis, and had target vessel revascularisation. We assumed that they also had a harmful event with regard to quality of life and angina, defined as the group mean plus both one and two SDs of the group mean (Jakobsen 2014).
'Worst‐best‐case' scenario: we assumed that all participants lost to follow‐up in the experimental group died, had a serious adverse event, had a major cardiovascular event, had stent thrombosis, and had target vessel revascularisation. We assumed that they also had a harmful event with regard to quality of life and angina, defined as the group mean plus both one and two SDs of the group mean (Jakobsen 2014). We assumed that all of those with missing outcomes in the control group survived, had no serious adverse event, had no major cardiovascular event, had no stent thrombosis, and had no target vessel revascularisation. We assumed that they also had a beneficial event with regard to quality of life and angina, defined as the group mean plus both one and two SDs of the group mean (Jakobsen 2014).

We presented results from both scenarios.

To assess the potential impact of missing SDs for continuous outcomes, we planned to perform the following sensitivity analyses.

Where SDs were missing and not possible to calculate, we would impute SDs from trials with similar populations and low risk of bias.
If we found no such trials, we would impute SDs from trials with a similar population. As the final option, we would impute SDs from all trials.

However, as we included no trials reporting data on continuous outcomes, we did not perform the above sensitivity analyses.

Results

Description of studies

Results of the search

We identified a total of 76,576 potentially relevant references through searching the CENTRAL (Cochrane Library) (n = 8778), MEDLINE (n = 21,652), Embase (n = 14,916), SCI‐EXPANDED (n = 19,920), BIOSIS (n = 10,657), and LILACS (n = 653) from inception to January 2017. The search strategy is presented in Appendix 2. Additionally, we found 18 trials by searching Google Scholar, ClinicalTrials.gov, and references identified through previous meta‐analyses and randomised clinical trials. We excluded 34,514 reference duplicates. Accordingly, we screened 42,080 records and excluded 41,985 references based on titles and abstracts. We assessed 95 full‐text articles for eligibility. Of these, we excluded 27 references reporting on 25 trials based on our inclusion and exclusion criteria. We have provided reasons for exclusion the Characteristics of excluded studies table. Fourteen trials/15 comparisons reported on a mix of acute coronary syndrome participants and stable angina participants and are awaiting classification (Characteristics of studies awaiting classification) (BASKET 2005a; BASKET 2005b; BASKET‐PROVE I 2010; BASKET‐PROVE II 2015; EAGLE 2006; Erglis 2007; FUTURE I 2004; FUTURE II 2004; Han 2007; Li 2004; NORSTENT 2016; PRODIGY 2014; SES‐SMART 2004; XIMA 2014; ZEUS 2015). We tried but could not establish contact with the trial authors to obtain individual participant data. We therefore did not include these trials in the primary analyses, but only in our exploratory analyses.

We included 51 publications reporting results of 25 trials (26 comparisons). The study flow chart is presented in Figure 1.

Figure 1

study flow diagram.

Included studies

We included 51 references reporting on 25 trials (26 comparisons) in our primary analyses comparing drug‐eluting stents versus bare‐metal stents. The trials were published between 2004 and 2016. The trials (which often took place in more than one country) were conducted at sites in 21 different countries: eight from Italy; eight from the Netherlands; five from Spain; three from Germany; three from the UK; three from Denmark; two each from Switzerland, Argentina, Austria, and Israel; and one each from Brazil, China, Croatia, Finland, France, Iran, Norway, Poland, Serbia, Sweden, and the USA. For further details on included studies please see Characteristics of included studies.

At one‐month follow‐up (median one‐month follow‐up), 6/25 trials reported all‐cause mortality, 7/25 trials reported serious adverse events, 6/25 trials reported major cardiovascular events, 5/25 trials reported cardiovascular mortality, 6/25 trials reported myocardial infarction, 5/25 trials reported stent thrombosis, and 6/25 trials reported target vessel revascularisation. No trials reported quality of life or angina.

At maximum follow‐up (median 12 months, range 1 to 72 months' follow‐up), 21/25 trials reported all‐cause mortality, 22/25 trials reported serious adverse events, 19/25 trials reported major cardiovascular events, 14/25 trials reported cardiovascular mortality, 18/25 trials reported myocardial infarction, 20/25 trials reported stent thrombosis, and 22/25 trials reported target vessel revascularisation. No trials reported quality of life or angina. Serious adverse events was not reported comprehensively and were limited to all‐cause mortality, cardiovascular mortality, myocardial infarction, stent thrombosis, and target vessel revascularisation.

Participants

The 25 included trials (26 comparisons) randomised a total of 12,503 participants with acute coronary syndrome. The number of participants in each trial ranged from 16 to 3006. The mean age was 60.4 years (range 55.9 years to 65.9 years) in the drug‐eluting stents group and 61.2 years (range 57.1 years to 64.6 years) in the bare‐metal stents group. The mean proportion of women was 21.1% in the drug‐eluting stents group and 31.3% in the bare‐metal stents group. In the drug‐eluting stents group 15.4% of participants had diabetes, whereas in the bare‐metal stents group 14.2% had diabetes. In both the drug‐eluting stents group and the bare‐metal stents group, 45% of participants had hypertension. In the drug‐eluting stents group 40.5% had hyperlipidaemia, whereas in the bare‐metal stents group 39.5% had hyperlipidaemia. The percentage of participants who had a previous myocardial infarction was 7.7% in the drug‐eluting stents group and 7.8% in the bare‐metal stents group. In the drug‐eluting stents group 6.6% had undergone a previous PCI, whereas in the bare‐metal stents group 5.7% had undergone a previous PCI. In the drug‐eluting stents group 2.3% had had a coronary bypass graft inserted, whereas in the bare‐metal stents group 1.9% had had a coronary bypass graft inserted.

In 22 trials (23 comparisons), the participants presented with STEMI. In two trials, the participants presented with different types of acute coronary syndrome. In one trial (ELISA‐3 2016), participants presented with NSTEMI.

In the drug‐eluting stents group, 50.1% had a lesion in the left anterior descending artery, 48.6% a lesion in the right circumflex artery, and 17.4% in the left circumflex artery. Under 1% had a lesion in the left main artery, intermediate branch, or in a graft. In the bare‐metal stents group, 49.8% had a lesion in the left anterior descending artery, 49.3% a lesion in the right circumflex artery, and 18.6% in the left circumflex artery. Under 1% had a lesion in the left main artery, intermediate branch, or in a graft.

In the drug‐eluting stents group, 62% of participants had one vessel disease and 38% had multivessel disease. In the bare‐metal stents group, 59% had one vessel disease and 41% had multivessel disease. In the drug‐eluting stents group, 62% of participants had a lesion with TIMI flow 0/1, 13.9% with TIMI 2, and 26.2% with TIMI 3. In the bare‐metal stents group, 61.6% had a lesion with TIMI flow 0/1, 13.9% with TIMI 2, and 26.3% with TIMI 3. Participants may have had a lesion in more than one vessel, hence the total of the above percentages is equal to more than 100%.

Experimental intervention

We included 11 trials where the drug‐eluting stent was a sirolimus‐eluting stent (with seven of these being a Cypher sirolimus‐eluting stent) and eight trials (nine comparisons) where it was a paclitaxel‐eluting stent (with six trials using a Taxus stent, either Express, Liberte, or unspecified). Two trials used an everolimus‐eluting stent, one trial a Biolimus‐eluting stent, one trial a dexamethasone‐eluting stent, one trial a zotarolimus‐eluting stent, and one trial a mix of drug‐eluting stents (sirolimus‐, paclitaxel‐ and zotarolimus‐eluting stents). In one trial it was unclear what type of drug‐eluting stent was used.

Successful PCI was performed in 95.2% of participants in the drug‐eluting stents group (reported by 16 trials). The average length of the drug‐eluting stent was 23.87 mm. The mean duration from onset of chest pain to PCI was 235.5 minutes (range 144 to 438 minutes) in the drug‐eluting stents group

Control intervention

The bare‐metal stent used in three trials was an Express bare‐metal stent, two trials a Driver stent, one trial an Express2 stent, one trial a Gazelle stent, one trial a Genius MAGIC stent, one trial a Multi‐Link Vision cobalt chromium stent, one trial a Vision stent, one trial a BiodivYsio stent, one trial a BX stent, and one trial a Liberte stent. In 10 trials the bare‐metal stent used was not specified. One trial used a mix of different bare‐metal stents.

Successful PCI was performed in 93.0% of participants in the bare‐metal stents group (reported by 16 trials). The average length of the bare‐metal stent was 22.29 mm. The mean duration from onset of chest pain to PCI was 242.2 minutes (range 164 to 426 minutes) in the bare‐metal stents group

Co‐interventions

All 25 trials (26 comparisons) employed co‐interventions. In 22 of these trials the study was designed to deliver the co‐interventions similarly in the drug‐eluting stents group and the bare‐metal stents group. In two trials the co‐interventions were not specified, and in one trial the drug‐eluting stents group and the bare‐metal stents group received different co‐interventions (different length of antithrombotic therapy). Before or at the time of PCI, aspirin was given to 99.8% of participants in both groups in the 10 trials that reported this. Clopidogrel was given to 85.4% of participants in the drug‐eluting stents group and 84.6% of participants in the bare‐metal stents group in the 10 trials that reported this. Heparin was given to 88.9% of participants in the drug‐eluting stents group and 89.8% of participants in the bare‐metal stents group in the eight trials that reported this. Glycoprotein IIb/IIIa receptor blocker was given to 63.5% of participants in the drug‐eluting stents group and 61.4% of participants in the bare‐metal stents group in the nine trials that reported this. For further details see Characteristics of included studies.

The post‐stent implantation therapy in 14 of the 25 trials was 80 mg to 100 mg of aspirin every day indefinitely. One trial administered 80 mg to 325 mg indefinitely; a second trial administered 250 mg for at least a year; and a third trial stated that participants received aspirin, but the dose was unclear. In eight trials it was unclear if the participants received aspirin in the months after stent implantation.

In three trials clopidogrel was administered for a year at an unspecified dose. In two trials 75 mg clopidogrel was administered for three months; in nine trials 75 mg was administered for nine to 12 months; and in four trials 75 mg was administered for six months. In five trials it was unclear if participants received clopidogrel in the months after stent implantation. One trial administered 500 mg ticlopidine for six weeks. Another trial administered 60 mg of prasugrel or 75 mg of clopidogrel for one year.

Excluded studies

Based on our inclusion and exclusion criteria, we excluded 27 publications on 25 trials after full‐text assessment. Seven trials did not assess drug‐eluting stents; four trials did not assess bare‐metal stents; one trial used a different medical intervention in the drug‐eluting stents group compared with the bare‐metal stents group; and 13 trials were not randomised clinical trials. For further details see Characteristics of excluded studies.

Risk of bias in included studies

Based on the information we collected from the published reports and received from authors, we judged all 25 trials (26 comparisons) to be at high risk of bias. We judged many trials to be at unclear risk of bias in several domains, and we were unable to obtain additional information from the authors when contacted. Additional information is provided in the 'Risk of bias' summary (Figure 2), 'Risk of bias' graph (Figure 3), and in the Characteristics of included studies table.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

The generation of the allocation sequence was at low risk of bias in 11 trials. The remaining 14 trials were described as being randomised, but the method used for sequence generation was not mentioned.

The method used to conceal allocation was at low risk of bias in 12 trials. In the remaining 13 trials the method used for allocation concealment was insufficiently described.

Blinding

The blinding of participants was at low risk of bias in one trial (ELISA‐3 2016). Nine trials stated that participants and personnel were not blinded. In the remaining 15 trials the method used to blind participants and personnel was insufficiently described.

The blinding of outcome assessors was performed and adequately described in 11 trials. One trial stated that outcome assessors were not blinded. In the remaining 13 trials the method used to blind outcome assessors was insufficiently described.

Incomplete outcome data

Thirteen trials adequately addressed incomplete outcome data. In the remaining 12 trials the method used to handle incomplete outcome data was insufficiently described.

Selective reporting

Four trials reported the results of the outcomes as specified in their respective protocols, or reported our primary outcomes, resulting in a judgement of low risk of bias according to our predefined 'Risk of bias' assessment. Five trials did not report the same outcomes as stated in the pre‐published protocol. For the remaining 16 trials, we could not obtain a protocol and the trial did not report our primary outcomes sufficiently.

Other potential sources of bias

Four trials had no other biases, resulting in a judgement of low risk of bias. Eleven trials were industry funded, resulting in an assessment of high risk of bias. The remaining 10 trials did not report how they were funded and were therefore judged to be at unclear risk of bias.