**Acute restoration of sinus rhythm in patients with paroxysmal atrial fibrillation: A Systematic Review and Network Meta-analysis** ***1.PICO*** •Population: Patients with paroxysmal AF •Intervention: Antiarrhythmic drug •Comparison: Antiarrhythmic drug or placebo •Outcome: Conversion to sinus rhythm in first 4 hours ***2.Methods*** This systematic review will follow the PRISMA guidelines (Liberati et al. BMJ 2009). ***2.1.Eligibility Criteria*** ***Inclusion criteria*** *Studies* • RCTs *Participants* •Study population >18 years old •Study population >50 patients •Patients with paroxysmal AF lasting less than 48 h and documented by electrocardiogram (Some AF paroxysms may continue for up to 7 days. AF episodes that are cardioverted within 7 days should be considered paroxysmal,2016 ESC Guidelines for the management of atrial fibrillation, doi:10.1093/eurheartj/ehw210) *Interventions* •Pharmacological cardioversion with intravenous or oral agent *Measurements* •Time was taken from the beginning of drug administration to conversion to sinus rhythm ***Exclusion criteria*** •Hemodynamic instability •Pregnancy •Atrial flutter •Sinus node disease •Acute coronary syndrome ***2.2.Outcome*** •Main: Conversion to sinus rhythm in first 4 hours •Secondary: Conversion to sinus rhythm in first 12 hours, Conversion to sinus rhythm in first 24 hours, Time was taken from the beginning of drug administration to conversion to sinus rhythm ***2.3.Search strategy*** •PubMed and Cochrane Central Register of Controlled Trials (CENTRAL) will be used for all searches. A search will be developed in clinicaltrials.gov for possible RCTs. A basic search strategy will be developed for PubMed and modified accordingly for other research engines. •We also searched Prospero to check if any similar systematic review is in progress in order to avoid duplication with ours. •Conference abstracts and references of relevant studies and systematic reviews will be perused, and experts will be contacted in order to identify unpublished studies. ***2.4.Data extraction, (selection and coding)*** Two main reviewers will screen the initial abstracts. If there is no agreement between the two reviewers regarding an abstract selection, then a third reviewer will review also this abstract and provide help on whether it should be included in data selection. The two main reviewers will read the full text articles and decide on whether they should be included in data extraction. Again, if there is no agreement between the two reviewers then a third reviewer will review also the full text article and provide help on whether the full text article should be included in data extraction. Two reviewers will independently extract the following data: Type of the study, number of patients in the study, observation period, intravenous or oral administration, clinical, laboratory and electrocardiographic variables, adverse events (death, sustained hypotension, bradycardia < 40 beats per minute, QT interval > 440 ms, ventricular arrhythmia, or any other event that required or prolonged hospitalization were considered serious adverse events) ***2.5.Subgroup analysis and investigation of heterogeneity*** *Potential subgroups for analysis will include:* • Sex • Age • Dose of drug • Intravenous or oral administration ***2.6.Data collection and analysis*** *Selection of studies* All studies will be imported in a reference management software (EndNote X7). After duplicate removal, two reviewers independently will screen all titles and abstracts and investigate full texts for eligible studies. Any differences about study eligibility will be resolved by discussion by a third review author. *Risk of bias (quality) assessment* Quality assessment for RCTs will be evaluated according to the revised Cochrane Risk of Bias tool(ROB) 2.0 by two independent reviewers. The tool consists of 5 different domains (randomization, deviations from intended interventions, missing outcome data, measurement of the outcome, selection of reported results). Every domain consists of unique questions. Each domain will be evaluated separately for each outcome. The overall risk of bias will be assessed as low, at some concerns or high according to the assessment of the 5 domains. Any disagreements will be solved by a third reviewer. ***2.7.Strategy for data synthesis*** For each outcome measure of interest, we will perform network meta-analyses (NMA) in the statistical software Rstudio. The method of NMA is an extension of the standard pairwise meta-analysis that enables a simultaneous comparison of multiple interventions, forming a connected network while preserving the internal randomization of individual trials. Frequentist approach will be adapted to fit the NMA model (Rücker, 2012). For continuous outcomes, we will calculate the mean difference (MD) and 95% confidence intervals (CI). Dichotomous outcomes will be analysed by calculating pooled odds ratio (OR) and 95% confidence intervals for each comparison. Furthermore, outcome‐specific network diagrams of the identified studies will be constructed. Treatments will be represented by nodes with size proportional to the number of participants randomized to each intervention and edges with width proportional to the number of studies evaluating each direct comparison. Moreover, we expect that transitivity assumption will be hold assuming that all pair-wise comparisons do not differ with respect to the distribution of effect modifiers (e.g., Dose of drug, Intravenous or oral administration). We will assess between-trial heterogeneity using Cochran’s Q test, the magnitude of heterogeneity parameter (tau squared) and I², with an I² value > 50% being considered as substantial heterogeneity. We will assume that heterogeneity will be constant across treatment contrasts (common tau squared). To check the assumption of inconsistency in the entire network we will use both global and local methods. We will assess the total inconsistency from the full design-by-treatment interaction random-effects model (Higgins et al. 2012). Also, the node-splitting approach (Dias et al Stat Med 2010) will be used to evaluate the inconsistency in our network, splitting our network estimates into the contribution of direct and indirect evidence, which allows us to control for inconsistency in specific comparisons. P-scores will be calculated in order to create a relative ranking of effectiveness for each outcome. We will use ‘comparison-adjusted’ funnel plots to assess the presence of small-study effects. In the presence of a small-study effect, contour-enhanced funnel plots will be checked to examine whether publication bias could be a reason. *Analysis of subgroups or subsets* The treatment effects for the primary outcomes will be explored in subgroup analysis for the following variables: Sex, Age, Dose of drug, Intravenous or oral administration Sensitivity analysis will be conducted in order to investigate the influence of studies that will be at a high risk of bias by removing them from the NMA. *Certainty of evidence* We will use the GRADE framework to rate the certainty of evidence for primary outcomes and classify evidence as high, moderate, low, or very low certainty.