2.1

Search strategy

A comprehensive search was conducted according to PRISMA statement using the following search terms: “Dual antiplatelet therapy, “DAPT”, “Antithrombotic therapy”, “triple therapy”, “anticoagulation”, “coronary artery disease”, “cardiovascular events” and “bleeding events”. Articles were retrieved from data base (MEDLINE, EMBASE and CENTRAL from inception to December 2016) and through manual screening of references of articles. Fig. 1 illustrates the selection process of included studies.

Search strategy using PRISMA statement.

2.2

Study selection

Inclusion criteria for studies are as follow:

• •
  

  Patients with PCI placed on DAPT and with a clear indication of OAC such as AF, prosthetic valves, VTE, LVT or aneurysm. Studies testing OAC for recurrent ischemia were excluded.

  
  
• •
  

  High quality studies (mean age ≥ 18 years, minimum sample size of ≥40 patients and follow up duration of >30 days) reporting outcomes of interests (see below) for participants who were randomized to either of the following groups: TT defined as combination of OAC and DAPT; DAPT defined as combination of two antiplatelet agents i.e. aspirin (ASA) and another anti platelet (P2Y12 inhibitors) or combination of OAC and SAP agent (OAC + SAP).

2.3

Quality assessment and data extraction

Data acquisition was done independently by authors (AA, SG and HR) using a standardized collection form incorporating information either directly or by manual calculation from the available variables.

Quality assessment for randomized controlled trials (RCTs) was done through Cochrane bias risk assessment tool (Supplement Table1) and the observational data were appraised by the Newcastle–Ottawa scale (NOS) Three main domains are assessed in NOS: selection of participants, comparability and outcome assessment. A score of six out of eight for a study on NOS was considered to be of high quality (Supplement Table 2).

2.4

Statistical analysis

The network analysis was carried out using winBUGS 1.4.3 (MRC Biostatistics Unit, Cambridge, UK) and NetMetaXL 1.6.1 (Canadian Agency for Drugs and Technologies in Health, Ottawa, Canada). All treatment groups were analyzed within mixed treatment comparison framework using Bayesian statistical methods Estimates were presented as OR with corresponding 95% CrI ranging from 2.5th to 97.5th centiles of posterior distribution. The relative ranking probability of being best for each treatment group was estimated through Markov chain Monte Carlo modeling. Results are arranged by ranking treatment in order of most pronounced impact on outcome measured based on SUCRA. A SUCRA of 90% depicts 90% effectiveness of treatment compared to other interventions; thus larger SUCRA values suggest better treatment. Heterogeneity variance with empirical distribution was derived based on publication by Turner et al. .

To assess the consistency, a traditional analysis was conducted between treatment arms using the Comprehensive Meta-analysis software version 3.0 (Biostat, Englewood, NJ). A p-value of 0.05 is set as significant. Heterogeneity was assessed using Q statistics with I ² > 50% being consistent with a high degree of heterogeneity. All the results have been reported according to random effects model. A sensitivity analysis for randomized and non-randomized studies was also conducted to assess the impact of randomization on outcomes.

2.5

Outcome measures

The primary safety outcome comprised of major bleeding events and minor bleeding events. The secondary effective endpoint was MACE: MI, stroke, and all-cause mortality. Other outcomes assessed were component of secondary endpoints along with CV mortality, TVR and stent thrombosis. Bleeding was assessed through Thrombolysis in Myocardial Infarction (TIMI) criteria where available. If the criteria were not met, a similar approach to Liu et al. was adopted where the definition was taken as reported in main article irrespective of minor differences. MI was defined as elevation of cardiac enzymes such as Troponin or CKMB irrespective of time of diagnosis. Stroke was defined as the new onset of neurological deficit secondary to ischemic etiology. Cardiovascular mortality was defined as death due to any cardiovascular causes whereas all-cause mortality included both cardiac and non-cardiac deaths.

2.1

Search strategy

A comprehensive search was conducted according to PRISMA statement using the following search terms: “Dual antiplatelet therapy, “DAPT”, “Antithrombotic therapy”, “triple therapy”, “anticoagulation”, “coronary artery disease”, “cardiovascular events” and “bleeding events”. Articles were retrieved from data base (MEDLINE, EMBASE and CENTRAL from inception to December 2016) and through manual screening of references of articles. Fig. 1 illustrates the selection process of included studies.

Search strategy using PRISMA statement.

2.2

Study selection

Inclusion criteria for studies are as follow:

• •
  

  Patients with PCI placed on DAPT and with a clear indication of OAC such as AF, prosthetic valves, VTE, LVT or aneurysm. Studies testing OAC for recurrent ischemia were excluded.

  
  
• •
  

  High quality studies (mean age ≥ 18 years, minimum sample size of ≥40 patients and follow up duration of >30 days) reporting outcomes of interests (see below) for participants who were randomized to either of the following groups: TT defined as combination of OAC and DAPT; DAPT defined as combination of two antiplatelet agents i.e. aspirin (ASA) and another anti platelet (P2Y12 inhibitors) or combination of OAC and SAP agent (OAC + SAP).

2.3

Quality assessment and data extraction

Data acquisition was done independently by authors (AA, SG and HR) using a standardized collection form incorporating information either directly or by manual calculation from the available variables.

Quality assessment for randomized controlled trials (RCTs) was done through Cochrane bias risk assessment tool (Supplement Table1) and the observational data were appraised by the Newcastle–Ottawa scale (NOS) Three main domains are assessed in NOS: selection of participants, comparability and outcome assessment. A score of six out of eight for a study on NOS was considered to be of high quality (Supplement Table 2).

2.4

Statistical analysis

The network analysis was carried out using winBUGS 1.4.3 (MRC Biostatistics Unit, Cambridge, UK) and NetMetaXL 1.6.1 (Canadian Agency for Drugs and Technologies in Health, Ottawa, Canada). All treatment groups were analyzed within mixed treatment comparison framework using Bayesian statistical methods Estimates were presented as OR with corresponding 95% CrI ranging from 2.5th to 97.5th centiles of posterior distribution. The relative ranking probability of being best for each treatment group was estimated through Markov chain Monte Carlo modeling. Results are arranged by ranking treatment in order of most pronounced impact on outcome measured based on SUCRA. A SUCRA of 90% depicts 90% effectiveness of treatment compared to other interventions; thus larger SUCRA values suggest better treatment. Heterogeneity variance with empirical distribution was derived based on publication by Turner et al. .

To assess the consistency, a traditional analysis was conducted between treatment arms using the Comprehensive Meta-analysis software version 3.0 (Biostat, Englewood, NJ). A p-value of 0.05 is set as significant. Heterogeneity was assessed using Q statistics with I ² > 50% being consistent with a high degree of heterogeneity. All the results have been reported according to random effects model. A sensitivity analysis for randomized and non-randomized studies was also conducted to assess the impact of randomization on outcomes.

2.5

Outcome measures

The primary safety outcome comprised of major bleeding events and minor bleeding events. The secondary effective endpoint was MACE: MI, stroke, and all-cause mortality. Other outcomes assessed were component of secondary endpoints along with CV mortality, TVR and stent thrombosis. Bleeding was assessed through Thrombolysis in Myocardial Infarction (TIMI) criteria where available. If the criteria were not met, a similar approach to Liu et al. was adopted where the definition was taken as reported in main article irrespective of minor differences. MI was defined as elevation of cardiac enzymes such as Troponin or CKMB irrespective of time of diagnosis. Stroke was defined as the new onset of neurological deficit secondary to ischemic etiology. Cardiovascular mortality was defined as death due to any cardiovascular causes whereas all-cause mortality included both cardiac and non-cardiac deaths.