Long‐term antithrombotic therapy for patients with atrial fibrillation and stable coronary artery disease

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the efficacy and safety of different long‐term antithrombotic regimens for people with AF and stable CAD.

Background

Description of the condition

Atrial fibrillation (AF) is a heart rhythm disorder that causes an irregular and often rapid heartbeat (Lip 2016). It is the most common sustained arrhythmia worldwide, affecting up to 33 million people globally (Chugh 2014). People with AF may have various symptoms. Palpitations, shortness of breath, and tiredness are the most frequent AF‐related symptoms (Hindricks 2020). An irregular heartbeat from AF, causes blood to pool in the atria, increasing the risk of blood clot formation. The clots can break off and be released into the circulatory system, leading to stroke (if clot blocks an artery in the brain) or systemic embolism (if clot blocks a peripheral artery). AF can increase the risk of stroke by five times, the risk of heart failure by three times, and the risk of death by 1.5 to 3.5 times (Hindricks 2020; Rahman 2014). AF increases in prevalence with advancing age and has been increasingly recognised as a major global health burden (Chugh 2014).

Coronary artery disease (CAD), also known as coronary heart disease, develops when the heart's blood supply is blocked or interrupted by a build‐up of fatty substances (plaque) in the coronary arteries (NHS 2020). CAD is the most common cardiovascular disease and remains the leading cause of cardiovascular morbidity and mortality globally (GBD 2018; Montalescot 2013). The most common symptom of CAD is angina (chest pain and discomfort). CAD is the main cause of heart attack (also known as myocardial infarction) (CDC 2021). Most heart attacks happen when a blood clot suddenly blocks the hearts' blood supply, causing heart damage. Over time, CAD can also weaken the heart muscle and may lead to heart failure or arrhythmias. People are considered to have stable CAD if they are asymptomatic or if their symptoms are controlled by medications or revascularisation.

Both diseases have similar, often coexisting risk factors, such as hypertension, diabetes mellitus, obesity and smoking (Hindricks 2020; Visseren 2021). People with AF and concomitant CAD are frequently encountered in daily practice; 33% of people with AF have CAD and up to 15% of people with stable CAD have concomitant AF (Goto 2008; Zoni‐Berisso 2014). CAD is associated with a high risk of AF within 30 days following a myocardial infarction, and the risk remains considerable over the subsequent five‐year period (Jabre 2011; Rahman 2014). AF combined with CAD is not only a common presentation in the clinical setting that makes antithrombotic therapy more complex, but it is also associated with significantly higher mortality (Lopes 2012; Ruff 2014a).

AF patients with risk factors for stroke and systemic embolism require long‐term anticoagulant therapy. Meanwhile, people with stable CAD require antiplatelet therapy to reduce the risk of ischaemic or atherothrombotic events, including stent thrombosis. Theoretically, people with AF and CAD may require a combination of both oral anticoagulation and antiplatelet therapy to prevent stroke and systemic embolism on the one hand, and coronary and vascular events on the other. However, such therapy also carries a greater risk of bleeding events, leading to hospitalisation or even death (ORBIT‐AF 2013). In this setting, it is crucial to identify the long‐term antithrombotic regimen with the optimal risk/benefit ratio, balancing the risk of bleeding, stroke, and ischaemia in this population.

Description of the intervention

The optimal long‐term antithrombotic therapy for patients with AF and stable CAD is unclear. Anticoagulants and antiplatelets are two classes of antithrombotic drugs and they prevent clot formation or thrombosis (Mega 2015). Anticoagulants work by inhibiting clotting factors to prevent the formation of blood clots, while antiplatelets work by preventing platelets from clumping and forming a clot (Mega 2015). Antithrombotic therapy usually consists of anticoagulant or antiplatelet agents, or a combination of these two modalities.

Long‐term oral anticoagulation is the standard treatment for people with AF who are at risk of stroke, and its use is guided by estimation of stroke risk according to the CHA₂DS₂‐VASc score (Kirchhof 2016). However, long‐term anticoagulation alone is not suitable for preventing new coronary events (Husted 2006; Knuuti 2020). The current options for oral anticoagulant (OAC) fall into two broad categories: vitamin K antagonists (VKAs), of which the most commonly used is warfarin; and non‐vitamin K antagonist oral anticoagulants (NOACs), also known as direct oral anticoagulants (DOACs), including the oral direct thrombin inhibitors (e.g. dabigatran) and the oral factor Xa inhibitors (e.g. rivaroxaban, apixaban and edoxaban) (Mega 2015).

VKAs were the first OACs used for AF. VKAs are very efficacious for prevention and treatment of thromboembolism, and have been shown to reduce stroke by up to 60% in AF patients, with a recommended international normalised ratio (INR) range of 2.0 to 3.0 (Hart 2007). In addition, VKAs have been used in people with CAD for the prevention of atherothrombotic events, showing effectiveness in reducing major cardiovascular outcomes (Hurlen 2002). However, the effectiveness of VKAs is dependent on the quality of anticoagulation control, with an INR time in therapeutic range (TTR) > 65% to 70% (Lip 2018). VKAs have a slow onset of action, it may take five to seven days before the INR is in the therapeutic range, with large individual variations. The use of VKAs is limited by their narrow therapeutic range, unpredictable dose response, and multiple food and drug interactions, which necessitate close monitoring of the INR and dose adjustments (Hirsh 2003).

NOACs are alternatives to VKAs for stroke prevention in AF, and have emerged as the first choice according to AF guidelines worldwide (Chong 2021; Hindricks 2020; January 2019). These medications have been shown to be at least as effective as warfarin for preventing stroke and systemic embolism in non‐valvular atrial fibrillation, and with more favourable safety profiles (Ruff 2014b). But NOACs are not recommended for AF with mechanical heart valves and moderate‐to‐severe mitral stenosis (Hindricks 2020). Compared to VKAs, NOACs offer convenient fixed dosing, rapid onset of action, fewer food and drug interactions, have an improved efficacy/safety ratio, and a predictable anticoagulant effect without the need for routine coagulation monitoring (Steffel 2018). NOACs are all at least partially reliant on renal elimination, therefore renal function needs to be monitored regularly and the dose adjusted accordingly (Steffel 2018). It should be noted that in studies using NOACs, inappropriate under/over dosing was associated with higher risk of stroke/systemic embolism, bleeding and death compared to on‐label dosing (Amarenco 2019; Camm 2020; Steinberg 2016).

Long‐term antiplatelet treatment is the standard recommended antithrombotic therapy in people with stable CAD, especially following coronary stenting (Knuuti 2020; Montalescot 2013). Antiplatelet agents such as aspirin and P2Y₁₂ inhibitors (if intolerant to aspirin) are cornerstone therapies for secondary prevention in CAD (Knuuti 2020). However, antiplatelet therapies (either single or dual antiplatelet therapy (DAPT)) are less effective than OACs in reducing the incidence of stroke associated with AF, whereas DAPT is associated with a bleeding risk similar to OAC therapy (Connolly 2006; Hart 2007). Hence, current guidelines do not recommend antiplatelet therapy alone for stroke prevention in AF, regardless of stroke risk (Hindricks 2020; January 2019).

People with both AF and CAD typically receive combination OAC and antiplatelet therapy. For AF patients with acute coronary syndrome or recent stent implantation, or both, within the last year, continuing combined anticoagulation and antiplatelet therapies are ideal up to 12 months, but for people with AF and stable CAD the evidence is limited (Lip 2019).

Several observational studies found that adding antiplatelet therapy to oral anticoagulation was associated with a higher risk of bleeding without a clear benefit on ischaemic end points in people with AF and stable CAD (Fischer 2018; Hamon 2014; Lamberts 2014; Lemesle 2017; Patti 2018). The European guidelines generally recommend OAC monotherapy (either VKA or NOAC) for long‐term therapy in people with both AF and stable CAD (Kirchhof 2016; Knuuti 2020; Lip 2019; Valgimigli 2018), and to only consider dual therapy with OAC and single antiplatelet agent (aspirin or clopidogrel) in people at high risk of coronary events. The CHEST antithrombotic therapy guidelines suggest OAC with either a NOAC or adjusted‐dose VKA therapy alone (Lip 2018), rather than the combination of OAC and aspirin (weak recommendation, low‐quality evidence). The North American consensus advises a full‐dose OAC monotherapy for long‐term therapy (Angiolillo 2018), and combination therapy consisting of OAC plus single antiplatelet therapy in people with high thrombotic and low‐bleeding risk.

However, these guidelines have created gaps between clinical practice and recommendations in treating people with AF and stable CAD due to the lack of supportive concrete evidence. There has been concern on the one hand that not receiving antiplatelet therapy could expose such people to higher rates of ischaemic events and on the other hand that dual therapy with full‐dose OACs and antiplatelet agents may lead to a higher risk of major bleeding events compared to OACs alone.

How the intervention might work

Anticoagulant and antiplatelet agents may reduce the incidence of thrombus formation based on their mechanism of action. However, these actions can also lead to bleeding as a major side effect of therapy.

VKAs interfere with the synthesis of functional coagulation factors (II, VII, IX, and X, as well as the anticoagulant proteins C and S) through inhibition of the vitamin K epoxide reductase (Ageno 2012). Among NOACs, there are direct thrombin inhibitors (e.g. dabigatran), which is a competitive, reversible direct inhibitor of the active site of both free and clot‐bound thrombin; and factor Xa inhibitors (e.g. rivaroxaban, apixaban, edoxaban), which are reversible direct inhibitors of free and clot‐bound factor Xa (Ageno 2012).

Antiplatelet drugs interfere with one or more steps in the process of platelet activation and aggregation, thereby significantly reducing the risk of thrombosis (Eikelboom 2012). Low‐dose aspirin acts through irreversible inhibition of platelet cyclooxygenase‐1 (COX‐1), reducing production of thromboxane A2 and thus diminishing platelet aggregation, thereby interfering with the formation of thrombi (Eikelboom 2012). The platelet P2Y₁₂ inhibitors (e.g. ticlopidine, clopidogrel, prasugrel, ticagrelor and cangrelor) block the binding of adenosine diphosphate to a specific platelet receptor P2Y₁₂, thereby inhibiting platelet activation, irreversible degranulation, shape change, and aggregation (Eikelboom 2012).

Why it is important to do this review

The optimal antithrombotic regimen for people with AF and stable CAD has not been well studied. The 2016 European AF guideline and some expert consensus recommend using oral anticoagulation alone without any antiplatelet therapy as the default strategy to limit the risk of bleeding in this population (Angiolillo 2018; Kirchhof 2016; Lip 2018; Lip 2019; Valgimigli 2018). However, these recommendations were mainly based on several observational studies with many biases, and the evidence supporting this strategy is sparse and weak, leading to uncertainty regarding this strategy. The question that remains is whether OAC monotherapy can be an effective alternative to antiplatelet therapy for stable CAD, or whether an antiplatelet agent should be added to OAC therapy for those people at high risk of ischaemic coronary events.

Two recent randomised controlled trials (RCTs) have investigated the role of OAC monotherapy, compared with combination OAC and single antiplatelet therapy in people with AF and stable CAD (AFIRE 2019; OAC‐ALONE 2019). The findings support for use of OAC alone among people with stable CAD who need long‐term anticoagulation. The recently updated European AF guideline and the North American consensus cited the findings of these trials as justification for recommending that using OAC alone as a long‐term antithrombotic strategy for people with AF and stable CAD (Angiolillo 2021; Hindricks 2020). However, OAC plus single antiplatelet therapy also seems to be reasonable for people with high thrombotic and low‐bleeding risk. In addition, both the AFIRE and OAC‐ALONE trials were conducted in Asian populations for whom the relative risks for thrombosis and bleeding may be different from some other populations. Therefore, it is necessary to perform a systematic review of all available trials of long‐term antithrombotic therapy for people with AF and stable CAD.

Objectives

To assess the efficacy and safety of different long‐term antithrombotic regimens for people with AF and stable CAD.

Methods

Criteria for considering studies for this review

Types of studies

We will include randomised controlled trials (RCTs), whether randomised at the level of the participant or as a cluster‐randomised design. We will not include cross‐over trials, due to the review focuses on the long‐term effect of the drugs being used and we will exclude observational studies. We will include studies reported as full‐text, those published as abstract only, and unpublished data.

Types of participants

We will include adult participants (18 years of age or older) who have a documented medical history of AF and stable CAD with a CHA₂DS₂‐VASc score of ≥ 1 in men or ≥ 2 in women.

AF includes paroxysmal, persistent, long‐standing persistent or permanent AF, and should be documented by a standard 12‐lead electrocardiogram (ECG), or as an episode lasting at least one minute on a rhythm strip, ambulatory ECG monitoring (e.g. Holter), or intracardiac electrogram (from an implanted pacemaker or defibrillator).

Stable CAD is defined as coronary artery stenosis ≥ 50% in one or several of the major coronary arteries (confirmed by coronary artery angiography or coronary computed tomography angiography) but not requiring revascularisation, or without an acute coronary syndrome in the previous 12 months, or revascularised CAD (either percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG)). In such patients, the most recent revascularisation should be performed 12 months prior in the case of acute coronary syndrome and six months prior in the case of stable CAD.

If we find trials that include some eligible participants as well as ineligible ones, we will contact the trialists for the data of the subgroup of interest. If that fails, we will only include the trial if a minimum of 80% of participants meet the eligibility criteria. We will document these decisions in the review and explore the impact of including studies with a subset of eligible participants through a sensitivity analysis.

Types of interventions

We will include trials that compared different long‐term antithrombotic therapy regiments (administered for at least 12 months). We will place no constraints on dosage and frequency.

Antithrombotic therapy includes the treatment regimens of (i) OAC monotherapy (either VKA or NOAC), or (ii) antiplatelet agents (single or dual antiplatelet therapy), or (iii) OAC plus antiplatelet agent(s). We will consider the following comparisons.

• OAC monotherapy versus OAC plus antiplatelet

• OAC monotherapy versus antiplatelet

• OAC plus antiplatelet versus antiplatelet

• NOAC versus VKA

Types of outcome measures

Reporting one or more of the outcomes listed here in the trial is not an inclusion criterion for the review. Where a published report does not appear to report one of these outcomes, we will access the trial protocol and contact the trial authors to ascertain whether the outcomes were measured but not reported. Relevant trials which measured these outcomes but did not report the data at all, or not in a usable format, will be included in the review as part of the narrative. For outcomes that could occur more than once in a participant during follow‐up, we plan to assess the number of participants with at least one event. Since we are interested in long‐term outcomes in this review, we will analyse outcomes at the longest available follow‐up.

Primary outcomes

• All‐cause death

• Ischaemic stroke (ischaemic with or without haemorrhagic transformation)

• Myocardial infarction

• Major bleeding, as defined using a definition of the International Society on Thrombosis and Haemostasis (ISTH), which is defined as acute clinically overt bleeding and satisfies one of the following criteria: i) fatal bleeding; ii) bleeding in a critical area or organ (such as intracranial, retroperitoneal, intraocular, intraspinal, intra‐articular, pericardial, intramuscular with compartment syndrome); iii) bleeding causing a reduction in haemoglobin of 2g/dL or more; iv) leading to transfusion of ≥ 2 units of whole blood or red cells (Schulman 2005)

Secondary outcomes

• Cardiovascular death

• Systemic embolism

• Proportion of participants with a stent who experience stent thrombosis

• Haemorrhagic stroke

• Clinically relevant non‐major bleeding, as defined using a definition of the ISTH, which is defined as any sign or symptom of haemorrhage not meeting the criteria for major bleeding, but meeting at least one of the following criteria: requiring medical intervention by a healthcare professional, leading to hospitalisation or increased level of care, prompting a face‐to‐face (i.e. not just a telephone or electronic communication) evaluation (Kaatz 2015)

• Minor bleeding: all overt bleeding events not meeting the criteria for either major or clinically relevant non‐major bleeding

• Health‐related quality of life, measured using any validated scale (e.g. European Quality of Life‐5 Dimensions Questionnaire (EQ‐5D), 36‐Item Short Form Health Survey (SF‐36))

Search methods for identification of studies

Electronic searches

We will identify trials through systematic searches of the following bibliographic databases.

• Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library

• MEDLINE (Ovid, from 1946 onwards)

• Embase (Ovid, from 1980 onwards)

• Web of Science Core Collection (Clarivate Analytics, from 1900 onwards)

We will adapt the preliminary search strategy for MEDLINE (Ovid) for use in the other databases (Appendix 1). We will apply the Cochrane sensitivity and precision‐maximising RCT filter to MEDLINE (Ovid) (Lefebvre 2021), and adaptations of it to the other databases, except CENTRAL.

We will also conduct a search of ClinicalTrials.gov (www.ClinicalTrials.gov) and the WHO International Clinical Trials Registry Platform (ICTRP) Search Portal (apps.who.int/trialsearch) for ongoing or unpublished trials.

We will search all databases from their inception to the present, and we will impose no restriction on language of publication or publication status.

We will not perform a separate search for adverse effects of interventions, but will consider adverse effects described in included studies only.

Searching other resources

We will check reference lists of all included studies and any relevant systematic reviews identified for additional references to trials. We will also examine any relevant retraction statements and errata for included studies. We will contact authors for missing data and ongoing trials and search relevant manufacturers' websites (e.g. www.Bayer.com, www.Pfizer.com, www.daiichisankyo.com, www.boehringer-ingelheim.com) for trial information. For any studies identified as eligible from clinical trial register records, we will search the trials registry number on PubMed for study publications.

Data collection and analysis

Selection of studies

Two review authors (SZ, FL) will independently screen titles and abstracts for inclusion of all potential studies we identify as a result of the search, and code them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. If there are any disagreements, a third review author (PX) will be asked to arbitrate. We will retrieve the full‐text study reports/publication and two review authors (SZ, FL) will independently screen the full‐text and identify studies for inclusion, and identify and record reasons for exclusion of the ineligible studies. We will resolve any disagreement through discussion or, if required, we will consult a third review author (PX). We will identify and exclude duplicates and collate multiple reports of the same study so that each study rather than each report is the unit of interest in the review. We will record the selection process in sufficient detail to complete a PRISMA flow diagram and 'Characteristics of excluded studies' table (Liberati 2009).

Data extraction and management

We will use a data collection form for study characteristics and outcome data which has been piloted on at least one study in the review. Two review authors (SZ, MT) will extract study characteristics from included studies. We will extract the following study characteristics.

• Methods: study design, total duration of study, number of study centres and location, study setting and date of publication

• Participants: number (N) randomised, N lost to follow‐up/withdrawn, N analysed, mean age, age range, gender, body mass index, smoking history, type of AF, history of stroke, history of myocardial infarction, history of percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG), hypertension, heart failure, diabetes mellitus, renal function, stroke risk (e.g. CHA₂DS₂‐VASc score), bleeding risk (e.g. HAS‐BLED score), 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: funding for trial, and notable conflicts of interest of trial authors

Two review authors (SZ, MT) will independently extract outcome data from included studies. We will resolve disagreements by consensus or by involving a third review author (PX). One review author (MT) will transfer data into the Review Manager 5 file (Review Manager 2020). A second review author (SZ) will double‐check that data are entered correctly by comparing the data presented in the systematic review with the data extraction form. A second review author (PX) will spot check study characteristics for accuracy against the trial report.

Assessment of risk of bias in included studies

Two review authors (SZ, FL) will independently assess risk of bias for each study using RoB 2 (Sterne 2019), outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021a). We plan to use the RoB 2 Excel tool (available on the risk of bias website) to implement RoB 2 assessment (RoB 2 Excel 2019). We will assess cluster trials with an additional domain from the archived version of the RoB 2 tool specific to the signalling questions (RoB 2 cluster‐randomized trials 2021), according to the guidance in Chapter 23 of the Handbook (Higgins 2021b). Any disagreements will be resolved by discussion or by involving another review author (PX). We will assess the risk of bias of specific results of a trial according to the following domains.

• Bias arising from the randomisation process

• Bias due to deviations from intended interventions

• Bias due to missing outcome data

• Bias in measurement of the outcome

• Bias in selection of the reported result

We will use the robvis tool to create weighted bar plots of the distribution of risk of bias judgements within each bias domain. We will use RevMan Web to generate traffic light plots of the domain‐level judgements for each outcome.

We will perform an intention‐to‐treat (ITT) analysis to estimate the effect of assignment to intervention, when assessing the bias due to deviations from the intended interventions (Higgins 2021a).

We will assess the risk of bias for the outcomes that will be included in our summary of findings table. We will use the signalling questions in the RoB 2 tool and rate each domain as 'low risk of bias', 'some concerns' or 'high risk of bias'. We will summarise the risk of bias judgements across different studies for each of the domains listed for each outcome. The overall risk of bias for the result is the least favourable assessment across the domains of bias. We will provide the consensus decisions for the signalling questions (i.e. detailed RoB 2 data) in the full text review and store these data as supplementary data or files.

When considering treatment effects, we will take into account the risk of bias for the studies that contribute to that outcome.

Measures of treatment effect

We will analyse dichotomous data as risk ratios (RRs) with 95% confidence intervals (CIs). Given the type of outcomes, we will assume hazard ratios (HRs) and odds ratios (ORs) to approximate the same measure as RRs. We will analyse continuous data as mean difference (MD) if each trial used the same scale or standardised mean difference (SMD) if they used different scales. We will interpret the effect size for SMDs according to Cohen's criteria: trivial (SMD < 0.2), small (0.2 ≤ SMD < 0.5), moderate (0.5 ≤ SMD < 0.8) or large (SMD ≥ 0.8) (Cohen 1988).

For continuous outcomes (quality of life), we will apply the accepted minimal clinical important difference (MCID) to aid the interpretation of the effect measures. The MCID for quality of life depends on the scale used. For example, the European Quality of Life‐5 Dimensions Questionnaire (EQ‐5D) is the most common instrument to value health outcomes. EQ‐5D assesses five health dimensions (with three levels of problems) and an overall health visual analog scale (EQ‐VAS), while 10 points is the MCID for the EQ‐VAS (Luo 2010).

We will narratively describe skewed data reported as medians and interquartile ranges.

Unit of analysis issues

We will include RCTs with parallel design and cluster‐RCTs. If we identify any cluster‐RCTs, we will use the recommendation from the Cochrane Handbook, and ensure that the effective sample size is calculated using the intracluster correlation coefficient (Higgins 2021c). We will not include cross‐over trials. For trials with multiple arms, if the baseline characteristics are similar among these groups and the comparisons are independent, we will include data from these trials. If a RCT has two intervention groups and a single comparator group, we will divide the number of participants in the comparator group into the number of eligible intervention groups in order to avoid double‐counting of the comparator when included in the same analysis.

Dealing with missing data

We will contact investigators or authors in order to verify key study characteristics and obtain missing numerical outcome data where possible. Where possible, we will use the RevMan calculator to calculate missing standard deviations using other data from the trial, such as CIs, based on methods outlined in the Cochrane Handbook (Higgins 2021d). Where this is not possible, and the missing data are thought to introduce serious bias, we will explore the impact of including such studies in the overall assessment of results by a sensitivity analysis.

Assessment of heterogeneity

We will first assess methodological and clinical heterogeneity with respect to the type of participants, interventions and outcomes in the included studies. Second, we will inspect forest plots visually to consider the direction and magnitude of effects and the degree of overlap between CIs. Third, we will evaluate statistical heterogeneity by using the Chi² test with P < 0.1 indicating significant heterogeneity, and we will use the I² statistic to quantify statistical heterogeneity.

In cases of no heterogeneity, we will perform a fixed‐effect meta‐analysis, whereas if we identify substantial heterogeneity (I² > 50%), we will report it and explore possible causes by prespecified subgroup analysis. If the source of heterogeneity cannot be explained, we will consider the following options: use a random‐effects model with appropriate cautious interpretation, or provide a narrative overview and not aggregate the studies at all.

Assessment of reporting biases

If we are able to pool more than 10 trials, we will create and examine a funnel plot to visually explore possible small study biases for the primary outcomes (Egger 1997).

Data synthesis

We will include all eligible studies in the primary analysis and to assess the potential effects of studies at high risk/some concerns in a sensitivity analysis. We will undertake meta‐analyses only where this is meaningful, that is, if the treatments, participants and the underlying clinical question are similar enough for pooling to make sense. We will use RevMan 5 to combine outcomes from individual trials when they are consistent on clinical grounds and have available outcome data (Review Manager 2020). In the absence of substantial heterogeneity (I² > 50%) and provided that there are sufficient trials, we will combine the results using a fixed‐effect model. If heterogeneity is substantial, we will undertake random‐effects meta‐analysis with appropriate cautious interpretation, or provide a narrative overview and not aggregate the studies at all (Deeks 2021). For dichotomous outcomes, we will use Mantel‐Haenszel methods to calculate pooled RRs; we will analyse continuous outcomes using an inverse variance method for pooling MD, and where the studies use different scales, we will use SMD (Deeks 2021). All data will be accompanied by the 95% CI.

Subgroup analysis and investigation of heterogeneity

We plan to carry out the following subgroup analyses for the primary outcomes.

• Age: elderly (≥ 75 years of age) versus non‐elderly (< 75 years of age)

• Types of anticoagulants

• Type of antiplatelet therapy (aspirin versus P2Y₁₂ inhibitors)

• Risk of stroke at baseline, using CHA₂DS₂‐VASc score (1, 2 to 3, ≥ 4)

• Risk of bleeding at baseline (high versus low), using HAS‐BLED score (high risk ≥ 3, low‐to‐moderate risk < 3)

• Previous PCI or CABG versus no previous PCI or CABG

• Diabetes at baseline versus no diabetes at baseline

• Renal failure at baseline (creatinine clearance < 30 mL/min, 30 to < 50 mL/min, ≥ 50 mL/min)

We will use the formal test for subgroup differences in Review Manager 5 (Review Manager 2020), and base our interpretation on this.

Sensitivity analysis

We plan to carry out the following sensitivity analyses, to test whether key methodological factors or decisions have affected the main result.

• Only including studies with an overall low risk of bias

• To check for the consistency in primary outcomes between different statistical models (fixed‐effect models and random‐effects models)

• Excluding studies with serious missing data (if outcome data are missing for more than 20% of the randomised participants)

Summary of findings and assessment of the certainty of the evidence

We will create a separate summary of findings table for each of the four comparisons, using the following outcomes.

• All‐cause death

• Ischaemic stroke

• Myocardial infarction

• Major bleeding

• Cardiovascular death

• Systemic embolism

• Proportion of participants with a stent who experience stent thrombosis

We will use the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the certainty of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes. We will use methods and recommendations described in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2021) using GRADEpro GDT software (GRADEpro GDT 2015). The overall RoB 2 judgement will be used to feed into the GRADE assessment. A rating of high certainty evidence can be achieved only when most evidence comes from studies that met the criteria for low risk of bias. The certainty of evidence might be downgraded by one level when most of the evidence comes from individual studies either with a crucial limitation for one item, or with some limitations for multiple items. We will justify all decisions to downgrade the certainty of the evidence using footnotes and we will make comments to aid reader's understanding of the review where necessary.

Two review authors (SZ, FL) working independently will make judgements about evidence certainty, with any disagreements resolved through discussion or involving a third review author (MS). Judgements will be justified, documented and incorporated into reporting of results for each outcome.

Appendices

Appendix 1. Preliminary MEDLINE (Ovid) search strategy

1 Atrial Fibrillation/ (56034)

2 ((atrial or auricular) adj5 fibril$).tw. (77235)

3 AF.tw. (43548)

4 1 or 2 or 3 (102835)

5 Coronary Disease/ (130910)

6 Coronary Artery Disease/ (62906)

7 ((coronary or heart) adj3 disease$).tw. (278032)

8 CAD.tw. (42235)

9 CHD.tw. (26315)

10 (isch?emi$ adj3 heart).tw. (46136)

11 IHD.tw. (5842)

12 stable coronary.tw. (4535)

13 (coronary adj2 atheroscleros$).tw. (10232)

14 (coronary adj2 arterioscleros$).tw. (839)

15 exp Myocardial Ischemia/ (432076)

16 Myocardial ischemia.tw. (27638)

17 Acute coronary syndrome.tw. (23629)

18 ACS.tw. (23786)

19 Angina.tw. (54299)

20 myocardial infarct$.tw. (202592)

21 heart infarct$.tw. (775)

22 Heart attack$.tw. (5771)

23 Coronary stenosis/ (11901)

24 (coronary adj3 stenos$).tw. (14558)

25 or/5‐24 (694584)

26 4 and 25 (16151)

27 (antithrombotic or anti‐thrombotic).tw. (20245)

28 exp Anticoagulants/ (222350)

29 (anticoagulation or anti‐coagulation).tw. (44104)

30 (anticoagulant$ or anti‐coagulant$).tw. (66662)

31 indirect thrombin inhibitor$.tw. (41)

32 exp Vitamin K/ai [Antagonists & Inhibitors] (2908)

33 (vitamin K antagonist$ or VKA or VKAs).tw. (6756)

34 warfarin/ (19592)

35 (((warfarin or coumadin or coumarin$ or phenprocoumon or acenocumarol or phenindione) and fluindione) or dicoumarol or ethyl biscoumacetate).tw. (763)

36 (ximelagatran or dabigatran or rivaroxaban or apixaban or edoxaban).tw. (10224)

37 (OAC or OACs).tw. (6276)

38 exp Factor Xa Inhibitors/ (7441)

39 Factor Xa Inhibitor*.tw. (2410)

40 oral direct thrombin inhibitor*.tw. (383)

41 (non‐vitamin K antagonist oral anticoagulant$ or new oral anticoagulant$ or NOAC or NOACs).tw. (3977)

42 (direct oral anticoagulant$ or DOAC or DOACs).tw. (3858)

43 exp Platelet aggregation inhibitors/ (124278)

44 (antiplatelet or anti‐platelet).tw. (32787)

45 (platelet$ adj5 inhibit$).tw. (25449)

46 thrombocyte aggregation inhibit$.tw. (89)

47 aspirin/ (45119)

48 (aspirin or acetylsalicylic acid or acetyl?salicylic acid$ or ASA).tw. (81765)

49 Thromboxane A2/ai [Antagonists & Inhibitors] (763)

50 thromboxane a2.tw. (7454)

51 exp Thienopyridines/ (11082)

52 Thienopyridine$.tw. (1422)

53 exp Purinergic P2Y Receptor Antagonists/ (12478)

54 Purinergic P2Y Receptor Antagonists.tw. (0)

55 exp Cyclooxygenase Inhibitors/ (129182)

56 Cyclooxygenase Inhibitor$.tw. (4850)

57 phosphodiesterase inhibit$.tw. (6624)

58 (ticlopidine or clopidogrel or prasugrel or cangrelor or cilostazol or dipyridamole or ealprostadil or picotamide or indobufen).tw. (24882)

59 or/27‐58 (546562)

60 26 and 59 (3783)

61 randomized controlled trial.pt. (514593)

62 controlled clinical trial.pt. (93873)

63 randomized.ab. (522812)

64 placebo.ab. (219198)

65 clinical trials as topic.sh. (193196)

66 randomly.ab. (366157)

67 trial.ti. (237215)

68 61 or 62 or 63 or 64 or 65 or 66 or 67 (1372185)

69 exp animals/ not humans.sh. (4741227)

70 68 not 69 (1267644)

71 60 and 70 (807)

Contributions of authors

Zhan Shipeng conceived, designed, co‐ordinated and drafted the protocol.

Liu Fang edited and advised on parts of the protocol and made an intellectual contribution to the protocol.

Xia Peiyuan edited and advised on parts of the protocol and made an intellectual contribution to the protocol.

Tang Min advised on parts of the protocol and made an intellectual contribution to the protocol.

Shu Maoqin advised on parts of the protocol, provided clinical knowledge and made an intellectual contribution to the protocol.

Zhang Zhihui advised on parts of the protocol, provided clinical knowledge and made an intellectual contribution to the protocol.

Wu Xiaojiao advised on parts of the protocol, provided methodological knowledge and made an intellectual contribution to the protocol.

Sources of support

Internal sources

• No sources of support provided

External sources

• NIHR, UK

  This project was supported by the National Institute for Health Research (NIHR) via Cochrane infrastructure funding to the Heart Group. The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, National Health Service (NHS) or the Department of Health and Social Care.

Declarations of interest

Zhan Shipeng: no conflict of interest

Liu Fang: no conflict of interest

Xia Peiyuan: no conflict of interest

Tang Min: no conflict of interest

Shu Maoqin: no conflict of interest

Zhang Zhihui: no conflict of interest

Wu Xiaojiao: no conflict of interest

New