2.1 Registration
This study was conducted according to the reporting guidelines of system review and mesh meta-analysis (PRISMA-NMA) and has been conducted in PROSPERO( https://www.crd.york.ac.uk/PROSPERO/#myprospero )Conduct prospective registration (ID: CRD42023420169).
2.2 Retrieval Strategy
Retrieve PubMed, EMbase, Cochrane Library, Web of Science, CNKI, Wanfang, VIP database, SinoMed. The search was conducted from the establishment of the database to July 3, 2022. Constructing retrieval strategies around PICOS tools: (1) Study subjects: Patients with atrial fibrillation; (2) Intervention measures: oral anticoagulants; (3) Control group: oral anticoagulants (different from the intervention group); (4) Outcome indicators: cognitive impairment; (5) Research type: Clinical research. Table 1 shows the retrieval strategy (PubMed).
Table 1
Search strategy on PubMed
#1 | Anticoagulants [MeSH Terms] |
#2 | (((((((Anticoagulants [Title/Abstract]) OR (Anticoagulant [Title/Abstract])) OR (Anticoagulation [Title/Abstract]) OR (Indirect Thrombin Inhibitors [Title/Abstract])) OR (Anticoagulating agent [Title/Abstract]) OR (Anticoagulative agent [Title/Abstract])) OR (Antichromobotic [Title/Abstract]) |
#3 | #1 OR # 2 |
#4 | (((((((Warfarin [MeSH Terms]) OR (Dabigatran [MeSH Terms])) OR (Rivaroxaban [MeSH Terms]) OR (apixaban [MeSH Terms]) OR (edoxaban [MeSH Terms])) OR (beta xaban [MeSH Terms]) OR (Aspirin [MeSH Terms]) OR (Clopidogrel [MeSH Terms]) OR (Receivers, Purinergic P2Y12 [MeSH Terms]) |
#5 | (((OR (Coumadine [Title/Abstract]) OR (Dabigatran [Title/Abstract]) OR (Pradaxa [Title/Abstract]) OR (Rivaroxaban [Title/Abstract]) OR (Xarelto [Title/bstract]) OR (apixaban [Title/bstract]) OR (Eliquis [Title/bstract]) OR (aboxoma [Title/bstract]) OR (eliques [Title/bstract]) OR (lunast [Title/bstract]) OR (edoxaban [Title/bstract]) OR (Savaysa [Title/bstract]) OR (lixia [Title/bstract]) OR (rocks [Title/bstract]) OR (beta ban [Title/Abstract]) OR (BEVYXXA [Title/Abstract]) OR (Aspirin [Title/Abstract]) OR (Acetylsalicylic acid [Title/Abstract]) OR (2- (Acetyloxy) benzoic acid [Title/Abstract]) OR (Acylpyrin [Title/Abstract]) OR (Aloxiprimum [Title/Abstract]) OR (Colfast [Title/Abstract]) OR (Dispril [Title/Abstract]) OR (Eastrin [Title/Abstract]) OR (Ecorin [Title/Abstract]) OR (Endosprint [Title] tle/Abstract) OR (Magnicyl [Title/Abstract]) OR (Micristin [Title/Abstract]) OR (Polypirin [Title/Abstract]) )OR (Polypiryna [Title/Abstract]) OR (Solprin [Title/Abstract]) OR (Solvesan [Title/Abstract]) OR (Zorprin [Title/Abstract]) OR (Acetysal [Title/Abstract]) OR (Clopidogrel [Title/Abstract]) OR (Discover [Title/Abstract]) OR (Plavix [Title/Abstract]) OR (Receivers, Purinergic P2Y12 [Title/Abstract]) OR (Purinergic P2Y12 Receivers [Title/Abstract]) OR (Purinoceptor P2Y12 [Title/Abstract]) OR (P2Y12 Purinoceptors [Title/Abstract]) OR (P2Y (ADP) Receiver [Title/Abstract]) OR (P2Y (T) Receiver [Title/Abstract]) OR (P2Y12 Purinoceptor [Title/Abstract]) OR (Purinoceptor P2T [Title/Abstract]) |
#6 | #4 OR # 5 |
#7 | #3 OR # 6 |
#8 | Atrial Fibrillation [MeSH Terms] |
#9 | (((((Atrial Fibrillation [Title/Abstract]) OR (Atrial Fibrillations [Title/Abstract]) OR (Auricular Fibrillation [Title/Abstract]) OR (Auricular Fibrillations [Title/Abstract]) OR (atrium filtration [Title/Abstract]) OR (heart filtration atrium [Title/Abstract]) |
#10 | #8 OR # 9 |
#11 | (Cognition) OR (Cognitions) OR (Cognitive) |
#12 | #7 AND # 10 AND # 11 |
2.3 Inclusion criteria
(1) Study subjects: Patients diagnosed with atrial fibrillation. There is no limit to race, age, gender, or course of disease. (2) Intervention measures: Oral anticoagulants (NOAC): Dabigatran, Apixaban, Edoxaban, Betrixaban, Vitamin K anticoagulants (VKAs), Receivers, Purinergic P2Y12, Avastin, Aspirin, Clopidogrel). (3) Control group: oral anticoagulants (NOAC): Dabigatran, apixaban, edoxaban, betrixaban; Vitamin K antagonists (VKAs), Receivers, Purinergic P2Y12, Avastin, Aspirin, Clopidogrel (different from the intervention group). (4) Outcome indicators: cognitive impairment (MMSE score, incidence rate of dementia).(5) Research type: Clinical research.
2.4 Exclusion criteria
(1) Research with incomplete data; (2) Non clinical studies, such as reviews, abstracts, letters, plans, etc; (3) Research on intervention measures that do not meet inclusion criteria.
2.5 Literature Screening
Use the literature management software Endnote to screen the literature. Two researchers (Ning WL and Wang SH) first screened duplicate literature, and then screened by reading the title and abstract to determine the included and excluded literature. Finally, read the entire text and further confirm inclusion. During this process, two researchers independently screened the literature, and finally compared the remaining literature. If it was the same, it was ultimately included. If it was different, it was discussed and resolved by a third researcher (Tang HQ).
2.6 Data Extraction
Two researchers (Ning WL and Wang SH) designed a data extraction table based on the information needed for the study, including author, year, study subjects, gender, sample size, intervention measures, follow-up time, outcome indicators, quality evaluation information, and other content. When there is inconsistency in the information extracted by two people, they should first discuss and resolve it with each other. If it still cannot be resolved, a decision should be made through discussion.
2.7 Quality evaluation
Two researchers (Ning WL and Wang SH) independently conducted a bias risk assessment. Randomized controlled trials were evaluated using the Cochrane Handbook 5.1.0 version of the tool ROB for assessing bias risk in randomized controlled trials [13]. It includes the following seven fields: (1) random sequence generation, (2) allocation concealment, (3) participant and subject blind methods; (4) Result evaluation blind method, (5) completeness of result data, (6) selective reporting, and (7) other biased sources. According to the number of components that may have high ROB, the experiment is divided into three levels: high risk (5 or more), moderate risk (3 or 4), and low risk (2 or less).
The cohort study used the Newcastle Ottawa Scale (NOS) scale for bias risk assessment [14]. This table represents the score with ★, with a maximum score of 9 ★, and 7–9 ★ are relatively high-quality articles. This study used this table, with several evaluation items assigned to each aspect. When the assigned items meet the requirements, they are represented by ★, with the highest comparability being 2 ★.
2.8 Statistical Analysis
In the study of correlation, all variables are countable variables, calculated using OR and 95% confidence interval (CI). Due to potential differences between studies, we chose a random effects model for analysis instead of a fixed effects model [15]. We use the gemtc package of R4.2.0 software and, according to the PRISMA NMA manual [16, 17], use Markov chain Monte Carlo simulation chains in a Bayesian framework for NMA aggregation and analysis. We will use the node method to quantify and prove the consistency between indirect and direct comparisons, calculated through the instructions in Stata software, if the p-value is > 0.05. Consistency check passed [18].
Use Stata software to present and describe network diagrams of different exercise interventions. In the generated network diagram, each node represents different motor interventions and control conditions, and the lines connecting the nodes represent direct positive comparisons between interventions. The size of each node and the width of the connecting lines are directly proportional to the number of studies [19].
Sort the efficacy of each indicator to obtain the surface under the cumulative ran king (SUCRA), and plot the probability ranking into a graph. Use a percentage to represent SUCRA. A larger percentage indicates that the intervention is more effective, and a value of 0 indicates that the intervention is completely ineffective. Although SUCRA can effectively be rephrased as a percentage of the effectiveness or acceptability of exercise interventions, these scores should be interpreted with caution unless there are actual clinical differences between interventions [20].